article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
the study of the four nucleon ( @xmath6 ) system is interesting from a number of different perspectives .
first of all , many reactions involving four nucleons , like @xmath7(@xmath8,@xmath9)@xmath10 , @xmath7(@xmath8,@xmath11)@xmath12 , or @xmath13 ( the @xmath14 process ) , are of extreme astrophysical interest , as they play important roles in solar models or in big - bang nucleosynthesis ( bbn ) .
the @xmath14 process , for instance , is the source of the highest energy neutrinos from the sun .
moreover , @xmath6 systems have become increasingly important as testing grounds for models of the nuclear force .
while not the nuclear - structure `` imbroglio '' of heavy nuclei , the @xmath15 system is the simplest system that presents the complexity thresholds and resonances that characterize nuclear systems , and therefore is a very good testing ground of modern few - body techniques @xcite .
similarly , since the @xmath6 bound state is ( to a very good approximation ) a @xmath16 state , it is a good `` laboratory '' for the study of the strange quark components of the nucleon via parity - violating electron - scattering experiments @xcite .
the theoretical description of @xmath17 systems still constitutes a challenging problem from the standpoint of nuclear few - body theory . only recently , with the near - constant increase in computing power and the development of new numerical methods , has the study of the @xmath18-particle bound state reached a satisfactory level of accuracy ; the @xmath6 bound - state has been calculated to a few tenths of kev @xcite .
the study of @xmath6 scattering states , on the other hand , is less satisfactorily developed .
the same increases in computational power , however , have opened the possibility for accurate calculations of the @xmath6 observables using realistic models for nucleon - nucleon ( @xmath4 ) and three - nucleon ( @xmath5 ) forces .
these calculations have been performed mainly by means of the faddeev - yakubovsky ( fy ) approach @xcite and the kohn variational principle @xcite . in spite of this progress
, some disagreements still exist between theoretical groups , as for example , in the calculation of the _ n_-@xmath10total cross section in the peak region ( around @xmath19 mev ) @xcite . to make matters worse , even in cases where the theoretical calculations agree , they are often strongly at variance with the experimental data . for example , the proton analyzing power @xmath1 in
p_-@xmath12elastic scattering is underestimated by theory at the peak of the angular distribution by about 40% @xcite .
other disagreements are discussed in refs .
@xcite .
the existing data in the literature for 4@xmath20 scattering are of lesser quality , when compared with the excellent and abundant data that exists for the @xmath4and @xmath5 ( scattering ) systems .
the intensely studied @xmath21 system is ( unfortunately ) very difficult to describe theoretically due to the presence of the bound state and many higher - energy resonances @xcite .
this leads us to investigate _ p_-@xmath12scattering states .
this paper will focus on _
p_-@xmath12elastic scattering at low energies , which is simpler than the @xmath21 system to investigate theoretically .
however , the situation for the closely related _
n_-@xmath10elastic scattering case is worth briefly discussing first .
the quantities of interest in _ n_-@xmath10zero - energy scattering are the zero - energy total cross section @xmath22 and the coherent scattering length @xmath23 .
experimentally , only the total cross section @xmath24 has been measured with high precision for a large range of energies .
the extrapolation of the measured @xmath24 to zero energy is straightforward and the value obtained is @xmath25 b @xcite .
the coherent scattering length has been measured by neutron - interferometry techniques .
the most recent value reported in the literature is @xmath26 fm @xcite .
an additional estimation of @xmath27 fm has been obtained from _
p_-@xmath12data by using an approximate coulomb - corrected r - matrix theory @xcite .
these values should be compared with those obtained theoretically : @xmath28 b and @xmath29 fm @xcite
. as already mentioned , at higher energies ( specifically in the `` peak '' region ) there exist sizable discrepancies between different theoretical predictions .
the @xmath24 calculations by fonseca @xcite and pfitzinger , hofmann , and hale @xcite are in good agreement with the experimental data , while the calculations by the grenoble @xcite and pisa groups @xcite are well below the data .
the origin of this disagreement is still not clear .
let us consider now the situation for _
p_-@xmath12elastic scattering .
the zero - energy quantities for this case are more difficult to evaluate .
approximate values of the triplet and singlet scattering lengths have been determined from effective range extrapolations @xcite to zero energy of data taken mostly above 1 mev , and therefore suffer large uncertainties @xcite .
this problem has been reconsidered recently by george and knutson @xcite ; this new phase - shift analysis ( psa ) gave two possible sets of scattering - length values , both of which are at variance with the theoretical estimates ( see the discussion in ref .
@xcite . ) the world database of existing _
p_-@xmath12scattering data can be divided in three energy regions .
there is a set of cross section and proton analyzing power @xmath1measurements at very low energies , from @xmath30 to @xmath31 mev @xcite .
another energy region is for @xmath32 to @xmath33 mev which includes the recent @xmath1measurements at @xmath34 and @xmath35 mev @xcite .
however , these @xmath1data are not very precise .
the cross - section data in this region are similarly imprecise and sparse in number .
they are also very old , being the very first published cross - section data for _ p_-@xmath12elastic scattering found in the literature @xcite .
the third group of measurements ( for @xmath36 mev ) includes differential cross sections , proton and @xmath12 analyzing power measurements , and spin correlation coefficients of good precision .
references for all these measurements can be found in ref .
@xcite .
the calculations performed so far for _ p_-@xmath12scattering have shown a glaring discrepancy between theory and experiment in the proton analyzing power @xcite .
this discrepancy is very similar to the well known `` @xmath1puzzle '' in scattering .
this is a fairly old problem , already reported almost 30 years ago @xcite in the case of and later confirmed also in the case @xcite .
all theoretical calculations based on realistic @xmath4potentials ( even including @xmath5 forces ) underestimate the measured nucleon vector analyzing power @xmath1by about @xmath37 % .
the same problem also occurs for the vector analyzing power of the deuteron @xmath38 @xcite , while the tensor analyzing powers @xmath39 , @xmath40 and @xmath41 are reasonably well described @xcite .
the inclusion of standard models of the @xmath5 force have little effect on calculations of these observables . to solve this puzzle , speculations about the deficiency of the @xmath4potentials in @xmath42 waves ( where the spectroscopic notation @xmath43 has been adopted )
have been suggested .
looking at @xmath4scattering data only , this possibility does not seem to be ruled out @xcite .
however , after taking theoretical constraints into account @xcite , this solution was considered unlikely , and has been confirmed by recent calculations which show that this puzzle is not solved even when new @xmath4potentials derived from effective field theory are used @xcite . consequently , attention has focused on exotic @xmath5 force terms not contemplated so far .
the best solution seems to be given by the inclusion of a spin - orbit @xmath5 force @xcite ; such a force has a negligible effect on the observables already well reproduced by standard @xmath5 potential models ( such as the binding energies , unpolarized cross sections and tensor analyzing powers ) but it is very effective for solving or at least for noticeably reducing the discrepancy .
other explanations have been proposed in refs . .
it should be noted that the current understanding of the 3n interaction is rather poor @xcite , and new terms in the @xmath5 interactions derived from chiral perturbation theory have very recently been proposed @xcite .
these new models have to be tested primarily in the @xmath5 system , but the @xmath6 system could also play an important role .
in fact , @xmath5 force effects are expected to be sizeable in the @xmath6 system @xcite ; in the scattering system @xmath5 force effects are small @xcite and masked in part by the contribution of the coulomb potential @xcite . moreover ,
the system is essentially a pure isospin @xmath44 state .
tests of the @xmath45 channel in any @xmath5 force can only be satisfactorily performed in a @xmath6 system such as _ p_-@xmath12 . seeking to explore potential three - nucleon force effects in a different system , and to investigate this new @xmath17 @xmath1puzzle , a series of proton- @xmath12 elastic scattering measurements
have been made .
angular distributions of the differential cross section @xmath0and proton analyzing power @xmath1have been measured at several energies below 5 mev ; the analyzing power experiments were performed with a gas - cell target and the @xmath0measurements were performed using a supersonic gas - jet target .
the nominal proton energy @xmath3at which each experiment was performed is listed in table [ table : exp : energies ] .
these energies were chosen to maximize the improvement in the low energy database , and because theoretical calculations are tractable in this region .
in addition , we present new theoretical calculations for the _ p_-@xmath12system .
they are performed via an expansion of the wave functions of the scattering states in terms of a hyperspherical harmonic ( hh ) basis and using the complex form of the kohn variational principle @xcite .
these new calculations reach a much higher degree of accuracy than those performed previously using the correlated hyperspherical harmonic ( chh ) functions @xcite . in this paper
we present calculations based on the argonne @xmath46 ( av18 ) @xcite @xmath4potential , which represents the @xmath4interaction in its full richness , with short - range repulsion , tensor and other non - central components and charge symmetry breaking terms .
the calculations are performed without and with the inclusion of the urbana ix ( uix ) @xmath5 force @xcite .
the remainder of this paper is organized as follows : in section [ sec : theory ] , a brief description of the hh technique is reported ; in section [ sec : exp ] , the experimental setup and methods are discussed ; the cross section @xmath0and proton analyzing power @xmath1measurements are compared with the theoretical results in section [ sec : compare ] ; finally , section [ sec : conclusions ] is devoted to the conclusions and broader context of the present work .
the wave function @xmath47 describing a _ p_-@xmath12scattering state with incoming orbital angular momentum @xmath48 and channel spin @xmath49 ( @xmath50 ) , total angular momentum @xmath51 , and parity @xmath52 can be written as @xmath53 where @xmath54 vanishes in the limit of large intercluster separations , and hence describes the system in the region where the particles are close to each other and their mutual interactions are strong . on the other hand , @xmath55 describes the relative motion of the two clusters in the asymptotic regions , where the
_ p_-@xmath12interaction is negligible . in the asymptotic region
the wave function @xmath56 reduces to @xmath57 , which must therefore be the appropriate asymptotic solution of the schrdinger equation .
@xmath57 can be decomposed as a linear combination of the following functions @xmath58_{s } \otimes y_{l}(\hat{\bf y}_i ) \bigr ] _
{ jj_z } \left ( f_l(y_i ) { \frac{g_{l}(\eta_c , qy_i)}{q y_i } } \pm \text{i } { \frac{f_l(\eta_c , qy_i)}{q y_i } } \right ) \ , \label{eq : psiom}\ ] ] where @xmath59 is the distance vector between the proton ( particle @xmath60 ) and @xmath12 ( particles @xmath61 ) , @xmath62 is the magnitude of the relative momentum between the two clusters , @xmath63 the spin state of particle @xmath60 , and @xmath64 is the @xmath65 wave function .
the total kinetic energy @xmath66 in the center of mass ( c.m . ) system and the proton kinetic energy @xmath3 in the laboratory system are @xmath67 where @xmath68 is the reduced mass ( @xmath69 is the nucleon mass . )
moreover , @xmath70 and @xmath71 are the regular and irregular coulomb functions , respectively , with @xmath72 . the function @xmath73^{2 l+1}$ ] in
( [ eq : psiom ] ) has been introduced to regularize @xmath71 at small @xmath74 , and @xmath75 as @xmath74 becomes large , thus not affecting the asymptotic behavior of @xmath56 .
note that for large values of @xmath76 , @xmath77\ , \ ] ] and therefore , @xmath78 ( @xmath79 ) describes in the asymptotic regions an outgoing ( incoming ) _ p_-@xmath12relative motion .
finally , @xmath80 \ , \label{eq : psia}\ ] ] where the parameters @xmath81 are the @xmath49-matrix elements which determine phase - shifts and ( for coupled channels ) mixing angles at the energy @xmath82 .
of course , the sum over @xmath83 and @xmath84 is over all values compatible with a given @xmath51 and parity .
in particular , the sum over @xmath83 is limited to include either even or odd values such that @xmath85 .
the `` core '' wave function @xmath86 is expanded using the hh basis . for four equal mass particles ,
a suitable choice of the jacobi vectors is @xmath87 where @xmath9 specifies a given permutation corresponding to the order @xmath60 , @xmath88 , @xmath89 and @xmath90 of the particles . by definition
, the permutation @xmath91 is chosen to correspond to the order @xmath92 , @xmath93 , @xmath94 and @xmath95 .
for a given choice of the jacobi vectors , the hyperspherical coordinates are given by the so - called hyperradius @xmath96 , defined by @xmath97 and by a set of angular variables which in the zernike and brinkman @xcite representation are i ) the polar angles @xmath98 of each jacobi vector , and ii ) the two additional `` hyperspherical '' angles @xmath99 and @xmath100 defined as @xmath101 where the @xmath102 are the moduli of the jacobi vector @xmath103 .
the set of angular variables @xmath104 , and @xmath100 is denoted hereafter as @xmath105 .
a generic hh function is written as @xmath106_{\lambda m } { \cal p}(\phi_{2p},\phi_{3p } ) \
, \label{eq : hh4}\ ] ] where @xmath107 where @xmath108 are jacobi polynomials and the coefficients @xmath109 are normalization factors .
the quantity @xmath110 is the so - called grand angular quantum number .
the hh functions are the eigenfunctions of the hyperangular part of the kinetic energy operator .
another important property of the hh functions is that @xmath111 are homogeneous polynomials of the particle coordinates of degree @xmath112 .
a set of antisymmetrical hyperangular - spin - isospin states of grand angular quantum number @xmath112 , total orbital angular momentum @xmath113 , total spin @xmath114 , total isospin @xmath115 , total angular momentum @xmath51 , and parity @xmath116 can be constructed as follows : @xmath117 where the sum is over the @xmath118 even permutations @xmath119 , and @xmath120_{s_a } s_k\bigr]_{s_b } s_m \biggr]_{\sigma } \biggr \}_{jj_z } \biggl [ \bigl[\bigl [ t_i t_j \bigr]_{t_a } t_k\bigr]_{t_b } t_m \biggr]_{tt_z}\ .
\label{eq : phi}\ ] ] here , @xmath121 is the hh state defined in ( [ eq : hh4 ] ) , and @xmath63 ( @xmath122 ) denotes the spin ( isospin ) function of particle @xmath60 .
the total orbital angular momentum @xmath113 of the hh function is coupled to the total spin @xmath114 to give a total angular momentum @xmath51 and parity @xmath123 . the integer index @xmath124 labels the possible choices of hyperangular , spin and isospin quantum numbers , namely @xmath125 that are compatible with the given values of @xmath112 , @xmath113 , @xmath114 , @xmath115 , @xmath51 and @xmath116 .
another important classification of the states is to group them in `` channels '' : states belonging to the same channel have the same values of angular ( @xmath126 ) , spin ( @xmath127 ) , and isospin ( @xmath128 ) quantum numbers but different values of @xmath129 , @xmath130 .
each state @xmath131 entering the expansion of the @xmath6 wavefunction must be antisymmetric under the exchange of any pair of particles .
consequently , it is necessary to consider states such that @xmath132 which is true when the condition @xmath133 is satisfied .
the number @xmath134 of antisymmetrical functions @xmath135 having given values of @xmath112 , @xmath113 , @xmath114 , @xmath115 , @xmath51 and @xmath116 but different combination of quantum numbers @xmath124 ( see eq .
( [ eq : mu ] ) ) is in general very large .
in addition to the degeneracy of the hh basis , the four spins ( isospins ) can be coupled in different ways to @xmath49 ( @xmath115 ) . however , many of the states @xmath135 , @xmath136 are linearly dependent . in the expansion of a @xmath6 wave function it is necessary to include the subset of linearly independent states only , whose number is fortunately significantly smaller than the corresponding value of @xmath134 .
the main problem is the computation of the matrix elements of the hamiltonian .
this task is considerably simplified by using the following transformation @xmath138 the coefficients @xmath139 have been obtained using the techniques described in ref . .
then the kinetic energy operator matrix elements are readily obtained analytically , and the @xmath4(@xmath5 ) potential matrix elements can be obtained by one ( three ) dimensional integrals .
the details are given in ref . .
the @xmath49-matrix elements @xmath140 and functions @xmath141 occurring in the expansion of @xmath86 are determined by making the functional @xmath142= { \cal s}^j_{ls , l^\prime s^\prime}(q ) -\frac{m_{n}}{\sqrt{6}\,{\rm i } } \left \langle\psi^{l^\prime s^\prime jj_z } _ { 1 + 3 } \left | h - e_3-\frac{q^2}{2 \mu } \right
| \psi^{lsjj_z}_{1 + 3}\right \rangle \label{eq : kohn}\ ] ] stationary with respect to variations in the @xmath143 and @xmath141 ( kohn variational principle ) . here
@xmath144 is the @xmath65 ground - state energy . by applying this principle , a set of second order differential equations for the functions
@xmath145 is obtained . by replacing the derivatives with finite differences ,
a linear system is obtained which can be solved using the lanczos algorithm .
this procedure , which allows for the solution of a large number of equations , is very similar to that outlined in the appendix of ref . and
it will not be repeated here .
the main difficulties of the application of the hh technique are the slow convergence of the basis with respect to the grand angular quantum number @xmath112 , and the ( still ) large number of linearly independent hh states with a given @xmath112 .
also a brute - force application of the method is not possible even with the most powerful computers available , so one has to select a suitable subset of states @xcite . in the present work ,
the hh states are first divided into _ classes _ depending on the value of @xmath146 , total spin @xmath114 , and @xmath129 , @xmath130 . in practice , hh states of low values of @xmath147 , @xmath148 , @xmath149 are first included . between them , those correlating only a particle pair
are included first ( i.e. those with @xmath150 ) , then those correlating three particles are added and so on .
the calculation begins by including in the expansion of the wave function the hh states of the first class c1 having grand angular quantum number @xmath151 and studying the convergence of a quantity of interest ( for example , the phase - shifts ) by increasing the value of @xmath152 .
once a satisfactory value of @xmath153 is reached , the states of the second class with @xmath154 are added in the expansion , keeping all the states of the class c1 with @xmath153 . then @xmath155 is increased until the desired convergence is achieved and so on .
let us consider , for example , the case @xmath156 , where there is only one @xmath157 channel in the sum over @xmath158 of eq .
( [ eq : psia ] ) , namely @xmath159 , @xmath160 . consequently , for this wave the s - matrix reduces to a value which is parametrized ( as usual ) as @xmath161 , where @xmath162 is known as the `` elasticity parameter '' .
note that in the application of the kohn principle given in ( [ eq : kohn ] ) the value @xmath163 is not guaranteed : it is achieved only when the corresponding internal part @xmath164 is well described by the hh basis .
we have used the value of @xmath162 as a check of the convergence of the hh expansion . in cases of poor convergence ,
the value of @xmath162 has been found to depend very much also on the choice of @xmath165 , the function used to regularize the coulomb function @xmath71 .
this function has been chosen to depend on a non - linear parameter @xmath166 , and thus a test of the convergence is performed by analyzing the dependence of @xmath162 on the parameter @xmath166 . at the beginning of the calculation , when the number of hh functions is not large enough to get convergence , @xmath162 is extremely dependent on the value of @xmath166 .
the phase shift @xmath167 , however , depends less critically on @xmath166 . by increasing the number of hh components in the internal wave function we observe that @xmath168 , and that the dependence on @xmath166 becomes negligible .
notice that the convergence rate has been found to depend on the value of @xmath166 ; some critical values of this parameter exist where the convergence can be very slow .
however , it is not difficult to find ranges of values of @xmath166 where the convergence is fast and smooth and the final results are independent of @xmath166 . since we are interested in the convergence of the hh function , we have chosen @xmath166 in one of the `` favorable '' regions , where the convergence is achieved in a smooth and fast way .
a detailed study on this subject will be reported elsewhere .
let us now briefly discuss the choice of the classes of hh states for the @xmath169 case .
note that , since the wave under consideration is of negative parity , only hh functions with odd values of @xmath112 and @xmath170 have to be considered .
moreover , we consider in this work only states with total isospin @xmath171 , as the effect of the states with @xmath172 should be negligible .
the criteria used to select the appropriate classes of hh functions require that the states with lowest @xmath173 be considered first .
a few of the channels considered in the calculation have been reported in table [ table : theory : chan0m1 ] .
the final choice of classes for the case @xmath169 is detailed below . as can be seen in table [ table : extra ] ,
@xmath175 is estimated to be rather small at @xmath176 and @xmath35 mev . however
, for the largest energy the convergence seems not to be completely under control and higher values of @xmath152 through @xmath177 should be employed . in any case
, we can see that the missing phase - shift is less than 2 % .
analogous problems have been found for the @xmath178 and @xmath179 states , whereas for the other states the convergence did not present any difficulty . since at the moment
the inclusion of a greater number of states would require a significant increase in computing time , we have preferred to use the extrapolation outlined above for obtaining estimates for the converged phase - shifts and mixing parameters for the @xmath169 , @xmath178 and @xmath179 states . to show the effect of the extrapolation on the observables , we present in fig .
[ fig : conv ] the results for four _ p_-@xmath12observables at @xmath180 mev and calculations using the av18 potential .
the dashed and thin solid curves have been obtained using the @xmath174 phase shift calculated with different values for @xmath181 .
more precisely , the dashed ( solid ) curves have been obtained using the value @xmath182 ( @xmath183 ) obtained with the choice of @xmath181 reported in the first ( sixth ) row of table [ table : extra ] .
the thick solid curve has been obtained using the extrapolated value @xmath184 .
the other phase - shift were taken with their final values ( in particular , for the @xmath178 and @xmath179 phase - shifts , we used the extrapolated values obtained using a similar procedure as described above ) .
the four observables considered in the figure are : the differential cross section @xmath0 , the proton analyzing power @xmath1 , the @xmath12 analyzing power @xmath185 , and the spin correlation coefficient @xmath186 .
as can be seen , there is a good convergence for @xmath0 , @xmath1 , and @xmath186 .
the observable @xmath185is more sensitive to @xmath167 , and the convergence is more critical . in any case , however , the thick solid curves are rather close to the thin solid curves , showing that the convergence has been nearly reached .
a similar analysis has been performed also for the @xmath178 and @xmath179 phase - shifts and mixing parameters , with similar findings .
therefore , we can conclude that the convergence of the hh expansion is sufficiently good to obtain nearly correct predictions for the _ p_-@xmath12observables and allowing therefore for meaningful comparisons between theory and experimental data .
the calculated phase shifts are in reasonable agreement with those calculated by other methods ( for example , those in ref . ) .
elastic scattering observables at @xmath180 mev calculated using different values for the @xmath174 phase - shift .
the dashed and solid curves were obtained using different truncations of the hh expansion corresponding to choices of @xmath187 reported in the first and sixth rows of table [ table : extra ] .
the thick solid curves have been obtained using the extrapolated value for this phase - shift , obtained as explained in the text .
the experimental data are from ref .
@xcite ( open circles ) and from the present work ( solid circles ) .
the potential used is av18.,width=480 ] the authors would like to thank w. tornow , a. fonseca , and t. b. clegg for very useful discussions ; t. v. daniels and m. s. boswell for their assistance in the data taking ; m. h. wood for assistance with the gas - jet target ; and j. d. dunham , e. p. carter , and especially r. m. oquinn for copious technical assistance .
this work was supported in part by the u.s .
department of energy under grant no .
de - fg02 - 97er41041 . | we present new accurate measurements of the differential cross section @xmath0 and the proton analyzing power @xmath1 for proton-@xmath2 elastic scattering at various energies .
a supersonic gas jet target has been employed to obtain these low energy cross section measurements .
the @xmath0 distributions have been measured at @xmath3 = 0.99 , 1.59 , 2.24 , 3.11 , and 4.02 mev .
full angular distributions of @xmath1 have been measured at @xmath3 = 1.60 , 2.25 , 3.13 , and 4.05 mev .
this set of high - precision data is compared to four - body variational calculations employing realistic nucleon - nucleon ( @xmath4 ) and three - nucleon ( @xmath5 ) interactions . for the unpolarized cross section
the agreement between the theoretical calculation and data is good when a @xmath5 potential is used .
the comparison between the calculated and measured proton analyzing powers reveals discrepancies of approximately 50% at the maximum of each distribution .
this is analogous to the existing `` @xmath1 puzzle '' known for the past 20 years in nucleon - deuteron elastic scattering . |
neutral hydrogen high - velocity clouds ( hvcs ) have been known to exist for nearly half a century @xcite , but their cosmological significance and role in the milky way and local group remain ambiguous .
hvcs span a wide range of sizes on the sky : from enormous systems like complex c , subtending an angular area of roughly @xmath2 @xcite , to subdegree - scale compact high - velocity clouds , some of which are unresolved even with large single - dish radio telescopes @xcite .
as their name suggests , the radial velocities of hvcs relative to the local standard of rest ( @xmath3kms@xmath4 ) are inconsistent with galactic rotation , implying they reside either in the extended galactic halo or at extragalactic ( local group ) distances .
hvcs appear to be a common feature around spiral galaxies , with m31 @xcite and m83 @xcite hosting what may be analogs to the largest hvcs seen near the galaxy .
numerous physical scenarios for the formation of hvcs have been proposed , including the cooling component of a supernova - driven `` galactic fountain , '' @xcite , inflows of neutral gas condensing from the local intergalactic medium @xcite , or the gaseous signatures of the `` missing '' dark matter subhalos around the galaxy @xcite .
oddly , unlike recently - discovered , very low - mass satellite galaxies which appear to harbor both stars and dark matter ( e.g. , * ? ? ?
* ) but little or no gas @xcite , searches for stellar populations associated with hvcs have only succeeded in placing upper limits @xcite .
similarly , aside from one case @xcite , hvcs appear to contain little or no molecular gas @xcite .
the detection of associated ionized gas ( up to ovi ; * ? ? ?
* ; * ? ? ?
* ) , but not the very high ions tracing @xmath5k gas @xcite , supports the notion that hvcs may be related to nonequilibrium cooling processes , either from the igm or a galactic fountain .
thus , although the most extended hvcs appear to be located relatively near the galaxy @xcite , they may nonetheless be connected to extragalactic phenomena .
complex c , for example , lies at a distance of @xmath6kpc @xcite and exhibits a low but inhomogeneous metallicity @xcite that may be indicative of a combination of infalling primordial gas and outflowing enriched material . while _
compact _ hvcs ( chvcs ) were initially hypothesized to be faint local group denizens @xcite , evidence is mounting that most of them simply represent the smallest fragments of the `` normal '' hvc population .
for instance , @xcite concluded that , due to their large inferred masses and inconsistent velocity widths , chvcs as a population are more likely to be associated with the galactic halo than the local group ; additionally , searches for similar clouds in nearby local group analogs indicate that chvcs are probably clustered within 90kpc of the galaxy , and massive intragroup clouds are relatively rare @xcite .
the lack of large clouds beyond 50 projected kpc from m31 @xcite further suggests that most of the hvcs around our analogous spiral galaxy do not lie far beyond the halo .
@xcite recently found that at least 35% of chvcs exhibit head - tail morphologies ( suggesting interaction with a diffuse warm - hot galactic halo medium ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and most of these lie near larger hvc complexes . in certain rare cases ,
though , chvcs which are _ not _ associated with larger - scale hi structure may be candidates for low - luminosity dwarf galaxies . the reported detection of dust in complex c by ( * ? ? ? * hereafter m05 ) provides a novel way to test the connection between hvcs and chvcs .
if chvcs are simply scaled - down versions of complex c at similar distances , their dust emission properties should be similar .
on the other hand , if certain chvcs differ fundamentally from the non - compact hvcs ( e.g. , in their metallicities or galactocentric distances ) , this would manifest itself through correspondingly different dust emission properties .
m05 inferred the dust content of complex c by cross - correlating the far - ir surface brightness in the _ spitzer extragalactic first - look survey _ with observations in the same field @xcite ; however , a subsequent analysis by @xcite using more sensitive arecibo measurements and iras fluxes did not confirm the m05 detection in complex c ( but , interestingly , detected dust in hvc complex m ) .
new longer - wavelength data from the satellite @xcite may have finally found dust in complex c , albeit at the @xmath7 level ( * ? ? ? * hereafter planck11 ) . in short ,
the prevalence of dust in hvcs is still essentially unknown , yet this information is crucial to our understanding of both hvcs themselves and their significance in the milky way and local group . with much smaller sizes and often high peak column densities , chvcs may allow the _ direct _ detection of dust emission ; such detections ( or strong upper limits thereupon ) can therefore provide insight to the nature of these mysterious objects . here
we present a targeted _ spitzer - mips _ search for dust emission in three chvcs : one with high - resolution radio data revealing several dense arcminute - scale knots , and two other high - column density chvcs from the parkes all - sky survey ( hipass ) catalog @xcite .
this paper is organized as follows : in section [ sec_data ] we discuss the chvc sample and the _ spitzer _ observations ; section [ sec_measure ] describes the flux measurements from the _ spitzer _ images , in section [ sec_discussion ] we discuss possible interpretations of the _ spitzer _ observations ( and caveats thereto ) , with the overall results briefly summarized in section [ sec_summary ] .
due to its extremely compact , `` clumpy '' morphology as revealed by the high - resolution radio maps of bw04 , and their conclusion that this is potentially a good candidate for a `` missing '' dark matter minihalo , we selected chvc289 as our primary follow - up target .
this object exhibits high peak column densities in all its five knots ( @xmath8@xmath9 ) in the high - resolution maps , even though its properties appear unremarkable in single - dish observations ( peak @xmath10@xmath9 as observed with effelsberg ) , having a relatively small total flux compared to other chvcs due to its small size . to complement this observation , we selected two other chvcs from the hipass catalog of @xcite : chvc @xmath11 and chvc @xmath12 ( hereafter referred to as chvc319 and chvc148 respectively ) .
these were specifically chosen because of their very high peak column densities averaged over the parkes beam ( @xmath13@xmath14 ) , and compact sizes ( semimajor axes less than 0.2 degrees ) .
in addition to the hipass column densities and sizes , higher - resolution observations of chvc148 were obtained by @xcite on the 100 m effelsberg single - dish telescope and by @xcite on the westerbork interferometer .
although the detailed ( arcminute - scale ) structures of these chvcs are unknown due to the lack of high - resolution imaging comparable to that obtained by bw04 on chvc289 , their hipass column densities are over four times higher than that of chvc289 , making them good candidates for objects with high - column density knots that may harbor dust .
all three of these chvcs were observed in cycle 3 with the _ spitzer - mips _
@xmath15 m , @xmath16 m , and @xmath1 m channels ( mips1 , mips2 , mips3 respectively ; see table [ tab_log ] for a summary of the observations ) , and the data were processed using standard techniques given in the _ spitzer _ data analysis cookbook .
because of its extreme compactness and precisely - constrained knot positions , chvc289 was observed with a single mips large - field aot in mips1 and mips3 . since half of the mips2 array is unusable , in this band we used two separate observations offset by 160 to ensure complete coverage of the chvc .
these observations were designed to provide a full 5 by 5 image of the central portion of chvc289 in all three bands , though due to an inoperative mips3 readout there are two narrow strips of missing data in the @xmath1 m image .
the nominal integration time per pixel in channels 1 , 2 , and 3 are 648 , 629 , and 52 seconds respectively .
the other two chvc positions are only known to within a few arcminutes and their angular sizes are larger ( albeit with significantly higher column densities than chvc289 ) ; we thus instead obtained mips scan maps of the regions surrounding these two chvcs , but to lower sensitivity since their high column densities should correspond to stronger dust emission .
the scan maps were designed to cover @xmath17 regions with @xmath18 array ( 80 ) offsets to fill in the gaps caused by the bad mips3 detector pixels ; this area is sufficient to fully cover chvc148 and chvc319 with at least 2 overscans in each band .
basic characteristics of these observations are also listed in table [ tab_log ] .
for chvc289 , we employ the known positions and sizes of the clumps measured by bw04 with the _ australia telescope compact array _ ( atca ) , shown overplotted on the mips images in figures [ fig_chvc289 ] and [ fig_chvc289_ch12 ] .
the central , highest - column density clump was resolved by atca with a diameter of 90 .
this is roughly @xmath19 times larger than the mips @xmath1 m point response function and therefore ( provided the dust and densities are correlated ) the mid - ir dust emission from this clump should similarly be resolved by all mips channels .
the four less - dense clumps are unresolved by atca ( with sizes less than 50 ) ; unfortunately , these fall either partly or entirely on the bad - pixel gaps in the mips3 image , so we only consider the dense central clump here .
since the emission is nominally extended rather than a point - source , we measure the flux of the central chvc289 clump in each of the mips images with a 90 diameter circular aperture ( i.e. , approximately equal to the size measured with atca ) . the uncertainty on the flux is determined with the `` empty aperture method , '' where the same flux measurement is repeated for many randomly placed apertures on source - free regions of each image ;
the resulting dispersion in the flux measurements represents the uncertainty in the background levels , while the median provides an estimate of the background levels in the image . in none of the three mips bands is the flux in the chvc289 cores significant ; we therefore compute @xmath20 upper limits in all bands assuming the measured uncertainties .
the small - scale structures and positions of chvc148 and chvc319 are less well - constrained : the estimated size of chvc319 in the hipass catalog is given as @xmath21 @xcite , while chvc148 has a fwhm of @xmath22 @xcite .
the uncertainty ellipses of these clouds take up a substantial fraction of the mips field ; thus , the empty aperture method can not be used to estimate flux uncertainties in these cases . note that
although @xcite report a compact core in chvc148 , this comprises only 4% of the total flux from this chvc with a peak column density of @xmath23@xmath14 , or roughly an order of magnitude less than the central concentration in chvc289 ; any dust emission from this core is therefore likely to be negligible .
the mips @xmath1 m images of both of these chvc regions are shown in figure [ fig_otherchvc ] , with the positions and beam sizes overplotted .
no extended @xmath1 m emission is evident within the chvc areas , but a number of point sources are visible , particularly in the chvc148 region .
we ran _ source extractor _
@xcite on the @xmath1 m image to detect point sources and accurately measure their positions , and checked the other two mips images for counterparts .
@xmath15 m sources are clearly visible at the position of all @xmath1 m objects in chvc148 , and a search of the _ _ nasa - ipac extragalactic database _ _ reveals that each of these sources lies within @xmath24 ( i.e. , roughly within the @xmath1 m point - response function ) of a relatively bright galaxy .
we therefore conclude that these compact mid - ir sources are background galaxies rather than dust clouds intrinsic to the chvcs , and both of these chvcs therefore lack significant extended or compact dust emission .
as noted in the previous section , we were only able to derive upper limits on the dust emission in chvc289 in all three bands . with previously - measured correlation coefficients between column density and dust emission
, we can predict the mid - ir surface brightnesses that the densest core of chvc289 _ should _ exhibit if its dust properties are similar to hvc complex c or intermediate - velocity clouds ( ivc ) . here
we employ measurements made by two independent groups : m05 in the _ spitzer extragalactic first look survey _ , and the `` draco '' field analyzed by planck11 , both of which cover small parts of complex c. these analyses claimed a significant correlation of mid - ir emission with in complex c , assuming a simple proportionality ( @xmath25 in m05 , @xmath26 in planck11 ; both define the emissivities @xmath27 and @xmath28 in units of mjy sr@xmath4 ( @xmath29 @xmath9)@xmath4 ) .
furthermore , in their fitting , both allowed for independent mid - ir emission from known local , ivc , and hvc velocity components .
planck11 only include one ivc component in their fits while m05 consider two ; for clarity , in our comparison to m05 , we only consider their `` ivc1 '' and `` hvc '' components . in figure [ fig_flux ]
we show these expected surface brightnesses for the central core of chvc289 , assuming the @xcite column density of @xmath30@xmath9 .
estimates for the measured planck11 _ iras _ and m05 _ iras_/_mips _
ir- correlations are shown as points .
additionally , we estimate modified blackbody spectra , where @xmath31 and @xmath32 are the free parameters and @xmath33 is the standard planck function . ] based on the 100 , 350 , 550 , and @xmath34 m measurements in the planck11 draco field ( table 2 ) .
these curves are normalized to the iras @xmath35 m flux , and the iras @xmath36 m point is not included in the fit since dust emission at high frequencies is dominated by nonequilibrium processes .
we convert our mid - ir flux limits for the central chvc289 core to surface brightness upper limits by assuming that the flux is distributed evenly over the 90 measurement aperture .
since this is comparable to the size ( fwhm ) of the emission itself , in essence we are comparing the predicted and measured _ average _ surface brightnesses in the central chvc289 core ; due to the effects of beam smearing and the lack of a @xmath1 m detection , this is a reasonable approximation . these derived upper limits for all three bands are overplotted in [ fig_flux ] .
although the two bluer mips upper limits ( @xmath15 m and @xmath16 m ) are more or less consistent with the predicted emission for complex c - like dust , the @xmath1 m upper limit is a factor of @xmath37 below the prediction from m05 , and a factor @xmath38 below the planck11 blackbody curve ; the two redder mips bands strongly rule out ivc - like emission ( and our @xmath15 m measurement appears inconsistent with the expected flux from m05 ) .
unfortunately , without a detection in any of the bands , we are unable to place limits on the dust temperature in chvc289 .
as alluded to in section 1 , the presence of dust in hvc complex c has been a somewhat contentious topic . in large part
this stems from the presence of multiple galactic components along the line of sight ( including local ism , one or more intermediate - velocity components , and complex c itself ) which vary in column density , velocity , and even composition across the enormous area spanned by complex c. mid - ir emission , by contrast , lacks the detailed velocity information of , so 2d spatial correlations must take into account contributions from each distinct component , with each potentially exhibiting a different dust fraction and temperature . as a result , there can be substantial degeneracies between the various measured parameters , and properly accounting for uncertainties is paramount .
m05 first reported a significant detection of dust ( with @xmath39 , nominally an @xmath40 result in this band ) over a @xmath41deg@xmath42 field , using a @xmath43 minimization technique ; both calibration and statistical uncertainties were included .
however , as pointed out by @xcite , spatial variations in the dust - to - gas ratio of the `` local '' component ( which by far dominates the ir emission ) can introduce spurious signals for weaker components like the hvc . since this possibility was not taken into account by m05 , and since their analysis does not account for position - to - position statistical variations ( like the empty aperture method we employ , or the `` displacement map '' technique described by @xcite ) , it appears likely that m05 s uncertainties are underestimated ( especially for the hvc component ) . given this , plus the fact that m05 and planck11 analyzed different parts of complex c , the discrepancies between some of the m05 and planck11 data points in figure [ fig_flux ] are not surprising .
in particular it is notable that the quoted errors on the @xmath36 m and @xmath35 m fluxes are comparable between planck11 and m05 , despite the factor @xmath41 larger angular area analyzed by planck11 . on the other hand
, planck11 do not include data at @xmath15 m , and at @xmath1 m we can only interpolate their modified blackbody fits .
due to their far wider spatial and spectral coverage , we take the planck11 as the most reliable comparison to our data ; however , for completeness we include comparisons to the m05 points as their measurements are more analogous to ours . at worst ,
if the m05 @xmath1 m point is taken as an upper limit , our data provide a nearly order - of - magnitude more stringent limit , and also lie a factor @xmath44 below the expectation from planck11 .
none of the chvcs studied here show evidence for the presence of dust in our deep _
spitzer - mips _ imaging , despite their relatively high column densities .
chvc289 in particular , with its strong central core resolved with atca , should have been confidently detected at the @xmath45 level in the @xmath1 m band ( and perhaps marginally at @xmath16 m ) if its dust temperature and dust - to - gas ratio are similar to hvc complex c. as shown in figure [ fig_flux ] , the @xmath1 m upper limit is incompatible with a similar origin , lying a factor of @xmath44 below the flux expected from the planck11 measurements .
even more striking is the discrepancy in the two redder bands between chvc289 and ivc1 .
intermediate - velocity clouds are thought to reside closer to the galactic disk , perhaps resulting from outflowing material due to star formation or other processes ; however , the limits on chvc289 are a factor of @xmath46 below the ivc emission level predicted by either m05 or planck11 .
the dust - gas correlation for ivc2 as reported by m05 is not shown in figure [ fig_flux ] , but lies in between complex c and ivc1 .
the extreme compactness of chvc289 , along with its lack of detected ir emission , suggest that it represents a different type of system than either complex c or the ivcs . however , with the current data we can not ascertain whether this difference is intrinsic to chvc289 , or simply due to a greater distance from the galactic plane ( hence weaker ambient radiation field and lower dust temperature / emissivity ) .
the upper limit to chvc289 s @xmath1 m surface brightness is a factor 4 below the expectations from m05 , who estimate a dust temperature in complex c of @xmath47k ( though the uncertainties on this quantity are likely large , as well ) .
assuming black body emission , this discrepancy in flux is consistent with chvc289 having a lower temperature , @xmath48k , which in turn means ( if the temperature difference depends entirely on galactocentric distance and the dust heating scales as @xmath49 ) that chvc289 could in principle be only marginally ( @xmath50% ) farther away than complex c s average distance of 10kpc @xcite .
an even smaller difference in distance could explain the discrepancy between chvc289 and the better - constrained planck11 measurements .
of course , this rough estimate assumes that these two clouds have identical dust properties aside from their temperatures ; in reality other variables are likely to be in play ( e.g. , dust distribution and dust - to - gas ratio ) .
a robust determination of the physical conditions in chvc289 thus evades us , since the current data can not directly constrain the dust temperature of this cloud .
similarly , no mid - ir emission is seen in either of the other two systems presented here , chvc148 and chvc319 . however , this result is weaker since the small - scale structures of these chvcs are essentially unknown .
these objects were initially chosen for _ mips _ follow - up due to their compact ( single - dish ) sizes and far higher peak column densities than chvc289 ; given that ultra - compact , high - density cores in chvc289 were found under interferometric scrutiny by bw04 , correspondingly higher - density ( or more numerous ) cores may also exist in chvc148 and chvc389 . nonetheless , neither obvious clumps nor diffuse haze in the mid - ir is visible within either of these systems .
@xcite did in fact find a `` core '' ( albeit with much lower resolution than the bw04 observations ) in chvc148 , but this core comprises only 4% of the total hvc flux .
higher - resolution observations of these systems would allow a more detailed comparison with the mips data ; however , since no sources are seen in the mid - ir , these comparisons will likely produce only weak upper limits ( as with chvc289 ) .
this analysis partly hinges on a comparison of the mid - ir emission in the chvcs to estimates of the dust content of complex c , which is still uncertain even with the planck11 data ; the primary reason for this is the weakness of the complex c dust emission and dominant foreground /dust components .
moreover , as m05 point out , despite the evidence that complex c contains dust , no molecular gas has been detected thus far ; either through h@xmath51 absorption @xcite or @xmath52co emission @xcite . since dust and molecular gas elsewhere in the galaxy are strongly correlated , this appears somewhat incongruous .
m05 suggest that the dust in complex c may instead be confined to cold , compact cores with the surrounding warmer gas comprising most of the ; however , our observations strongly rule out such emission in chvc289 s compact cores . on the other hand , this may further highlight that chvc289 and complex c are different in some elementary way besides their angular sizes .
although the central clump of chvc289 is resolved with atca , it is only marginally so : the inferred clump size is 90 , compared to the @xmath53 beam reported by @xcite .
the morphology of the clump could actually be more complex , consisting of several denser clumps spread over the 90area .
since we assumed constant column density and dust emission over this area , the presence of such clumps could `` hide '' emission in higher ( but still undetected ) surface brightness regions , implying that our upper limits could in principle be too low .
however , such effects are unlikely to account for the large ( factor of @xmath44 at @xmath1 m ) discrepancy between the chvc289 and complex c emission properties , especially given the large column density measured over the same 90 area . since only upper limits in all three bands
can be calculated for chvc289 , it is not possible to compare the shape of the putative dust sed with the other systems ( complex c and the ivcs ) .
if chvc289 has a substantially flatter dust sed ( e.g. , due to a different grain size distribution or higher temperature ) , this may also account for its lower @xmath1 m flux . on the other hand ,
such scenarios would be necessarily contrived ( e.g. , chvc289 would need to be both hotter and very dust poor ) , and the overall conclusion would remain : it is a fundamentally different system than the ones studied by m05 and planck11 .
using deep _ spitzer - mips _ observations of three chvcs , we have placed limits on the dust emission from these enigmatic objects , ruling out similar physical conditions in hvc complex c and the most compact object in our sample ( chvc289 ) .
notably , the @xmath1 m surface brightness limit falls a factor of 2 below that inferred via _ planck _ measurements for dust in hvc complex c , and are even more discrepant with ivc measurements , indicating that chvc289 is either farther away or exhibits a lower dust - to - gas ratio than these other two object categories .
chvc289 ( and by extension other chvcs ) may therefore reside farther from the galactic disk or exhibit more `` primordial '' compositions .
alternatively , if all ( c)hvcs exhibit similar physical conditions and the m05/planck11 dust detections in complex c are spurious , our chvc289 measurement places the most stringent upper limit on hvc dust yet . unfortunately , since all we have are upper limits , the current data do not allow us to distinguish between these possibilities .
no extended dust emission is seen toward the other two chvcs ( chvc148 and chvc389 ) observed with _
mips _ ; however , their small - scale distributions are poorly constrained . since their total column densities are
much higher than chvc289 with comparable radii measured from single - dish observations , high - resolution maps of these two clouds might allow stronger constraints to be placed on their dust emission with the existing _ mips _ data .
a robust study of the dust content over the full chvc population ( and comparison to large - scale hvcs ) will require of a much larger sample than the three presented here .
in particular , the mission @xcite , with its full - sky coverage , sensitivity to cold dust at submillimeter wavelengths , and angular resolution well - matched to chvc sizes , will prove integral to this aim .
we are grateful to the anonymous referee for comments which substantially improved the manuscript , and to the _ spitzer _ team for their efforts with this remarkable instrument .
financial support for this work was provided by nasa through award number 1289595 issued by jpl / caltech .
. acknowledges additional support from the netherlands organization for scientific research ( nwo ) , the leids kerkhoven - bosscha fonds , and nsf grant ast-0707417 . this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
akeson , r. l. , & blitz , l. 1999 , , 523 , 163 bertin , e. , & arnouts , s. 1996 , a&as , 117 , 393 booth , c. m. , & theuns , t. 2007 , , 381 , l89 braun , r. , & burton , w. b. 1999 , , 341 , 437 bregman , j. n. 1980 , , 236 , 577 bregman , j. n. , otte , b. , irwin , j. a. , et al .
2009 , , 699 , 1765 brns , c. , & westmeier , t. 2004 , , 426 , l9 ( bw04 ) chynoweth , k. m. , langston , g. i. , holley - bockelmann , k. , & lockman , f. j. 2009 , , 138 , 287 collins , j. a. , shull , j. m. , & giroux , m. l. 2007 , , 657 , 271 combes , f. , & charmandaris , v. 2000 , , 357 , 75 de heij , v. , braun , r. , & burton , w. b. 2002 , , 391 , 67 dessauges - zavadsky , m. , combes , f. , & pfenniger , d. 2007 , , 473 , 863 gibson , b. k. , giroux , m. l. , penton , s. v. , et al .
2001 , , 122 , 3280 giovanelli , r. , haynes , m. p. , kent , b. r. , & adams , e. a. k. 2010 , , 708 , l22 grcevich , j. , & putman , m. e. 2009 , , 696 , 385 hopp , u. , schulte - ladbeck , r. e. , & kerp , j. 2007 , , 374 , 1164 kere , d. , & hernquist , l. 2009 , , 700 , l1 kwak , k. , shelton , r. l. , & raley , e. a. 2009 , , 699 , 1775 lockman , f. j. & condon , j. j. 2005 , , 129 , 1968 lockman , f. j. , benjamin , r. a. , heroux , a. j. , & langston , g. i. 2008 , , 679 , l21 maloney , p. , & putman , m. e. 2003 , , 589 , 270 miller , e. d. , bregman , j. n. , & wakker , b. p. 2009
, , 692 , 470 miville - deschnes , m .- a . , boulanger , f. , reach , w. t. , & noriega - crespo , a. 2005 , , 631 , l57 ( m05 ) moore , b. , ghigna , s. , governato , f. , et al . 1999 , , 524 , l19 muller , c. a. , oort , j. h. , & raimond , e. 1963 , cr acad .
paris , 257 , 1661 murphy , e. m. , sembach , k. r. , gibson , b. k. , et al .
2000 , , 538 , l35 peek , j. e. g. , heiles , c. , putman , m. e. , & douglas , k. 2009 , , 692 , 827 pisano , d. j. , barnes , d. g. , gibson , b. k. , et al .
2004 , , 610 , l17 pisano , d. j. , barnes , d. g. , gibson , b. k. , et al .
2007 , , 662 , 959 planck collaboration 2011a , , 536 , a1 planck collaboration 2011b , , 536 , a24 ( planck11 ) putman , m. e. 2006 , , 645 , 1164 putman , m. e. , de heij , v. , stavely - smith , l. , et al .
2002 , , 123 , 873 putman , m. e. , saul , d. r. , & mets , e. 2011 , , in press richter , p. , de boer , k. s. , widmann , h. , et al .
1999 , nature , 402 , 386 richter , p. , sembach , k. r. , wakker , b. p. , & savage , b. d. 2001 , , 562 , l181 sembach , k. r. , wakker , b. p. , savage , b. d. , et al .
2003 , , 146 , 165 shull , j. m. , jones , j. r. , danforth , c. w. , & collins , j. a. 2009 , , 699 , 754 siegel , m. h. , majewski , s. r. , gallart , c. , et al .
2005 , , 623 , 181 simon , j. d. , blitz , l. , cole , a. a. , weinberg , m. d. , & cohen , m. 2006 , , 640 , 270 smoker , j. v. , fox , a. j. , & keenan , f. p. 2011
, , 415 , 1105 sommer - larsen , j. 2006 , , 644 , l1 strigari , l. e. , bullock , j. s. , kaplinghat , m. , et al .
2008 , nature , 454 , 1096 thilker , d. a. , braun , r. , walterbos , r. a. m. , et al .
2004 , , 601 , l39 thom , c. , peek , j. e. g. , putman , m. e. , et al .
2008 , 684 , 364 tripp , t. , wakker , b. p. , jenkins , e. b. , et al .
2003 , , 125 , 3122 wakker , b. p. , murphy , e. m. , van woerden , h. , & dame , t. m. 1997 , , 488 , 216 wakker , b. p. , & van woerden , h. 1991 , , 250 , 509 wakker , b. p. , york , d. g. , howk , j. c. , et al .
2007 , , 670 , l113 wakker , b. p. , york , d. g. , wilhelm , r. , et al .
2008 , , 672 , 298 westmeier , t. , brns , c. , & kerp , j. 2005 , , 432 , 937 westmeier , t. , brns , c. , & kerp , j. 2008 , , 390 , 1691 williams , r. j. , mathur , s. , & nicastro , f. 2006 , , 645 , 179 williams , r. j. , mathur , s. , nicastro , f. , & elvis , m. 2007 , , 665 , 247 willman , b. , dalcanton , j. , ivezi , . , schneider , d. p. , & york , d. g. 2002 , , 124 , 2600 chvc289 & @xmath54 & @xmath55 & 648 & 629 & 52 & 0.073 & 0.45 & 1.1 & single pointing + chvc148 & @xmath56 & @xmath57 & 168 & 84 & 17 & 0.18 & 0.39 & 0.54 & mips scan map + chvc319 & @xmath58 & @xmath59 & 168 & 84 & 17 & 0.068 & 0.35 & 0.38 & mips scan map | we employ three - band spitzer - mips observations to search for cold dust emission in three neutral hydrogen compact high - velocity clouds ( chvcs ) in the vicinity of the milky way .
far - infrared emission correlated with column density was previously reported in hvc complex c , indicating that this object contains dust heated by the galactic radiation field at its distance of @xmath0kpc .
assuming published spitzer , iras , and planck , ir - correlations for complex c , our _ spitzer _ observations are of sufficient depth to directly detect @xmath1 m dust emission in the chvcs if it is present at the same level as in complex c , but no emission is detected in any of the targets .
for one of the targets ( chvc289 ) which has well - localized clumps , we therefore conclude that it is fundamentally different from complex c , with either a lower dust - to - gas ratio or a greater distance from the galactic disk ( and consequently cooler dust temperature ) .
firm conclusions can not be drawn for the other two _ spitzer_-observed chvcs since their small - scale structures are not sufficiently well known ; nonetheless , no extended dust emission is apparent despite their relatively high column densities .
the lack of dust emission in chvc289 suggests that at least some compact high - velocity clouds objects may exhibit very low dust - to - gas ratios and/or greater galactocentric distances than large hvc complexes . |
the use of exactly solvable models and methods leading to analytical solutions has always been important in physics in general .
models of particular interest are those that display some physical correspondence to the physics of quantum chromodynamics ( qcd ) or that may give insights to the difficult task , given its nonperturbative nature , of analyzing the phase structure at finite temperatures and densities .
this is also the kind of situation , regarding phase transitions in a hot and dense medium , that is expected to be realized in the early universe and in the recent experiments involving heavy - ion collisions .
the use of appropriate techniques and models in the context of quantum field theory that can increase our knowledge on this kind of phenomena are then timely and important .
one such phenomenon of topical interest is the study of the chiral symmetry restoration ( csr ) in qcd .
however , due to the intrinsic difficulty of approaching directly this study in the context of qcd itself , other more tractable models have been extensively used . among them ,
the simplest fermionic model displaying many properties common to qcd is the gross - neveu ( gn ) model @xcite .
for instance , in 2 + 1 dimensions it can exhibit a discrete spontaneous chiral symmetry breaking ( csb ) that happens for a critical coupling at zero temperature and density , while its spectrum can contain excitations like baryons as well as mesons ( which are composite fermionic states ) . even though it is a quartic interacting fermion model , thus nonrenormalizable ( in 2 + 1 dimensions ) in ordinary perturbation theory , it is renormalizable in a @xmath1 expansion ( where @xmath2 denotes the number of fermion species ) and it is exactly soluble in the large-@xmath2 limit @xcite .
these properties make this effective model one of the simplest and most useful frameworks for understanding some of the important physical features displayed by qcd . for the reasons above
, the gn model is a perfect laboratory for studying symmetry aspects in a hot and dense medium and outside the qcd realm , the model by its own and its extensions may also have applications in condensed matter physics in the understanding of the properties of fermion liquids and high-@xmath3 superconductivity models @xcite . besides being studied analytically mostly with the large-@xmath2 expansion technique , there are a large number of works studying the gn model numerically from the lattice monte - carlo perspective @xcite ( for an introduction to lattice qcd , its problems in implementation and applications to effective models , like the gn one , see @xcite and references therein ) . among the interesting thermodynamical properties of the gn model demonstrated by the large-@xmath2 technique
is the existence of a line of second order csr in the temperature and chemical potential ( @xmath4 ) plane , starting at a value ( @xmath5 ) and ending , abruptly , at a first order transition point ( @xmath6 ) @xcite . while this phase structure was verified by lattice simulations with finite number of fermion species , @xmath2 , in ref .
@xcite some indication was given for the existence of a first order transition line starting at a tricritical point ( @xmath7 ) and ending at @xmath8 .
however , the precise position of this tricritical point in the phase diagram could not be determined .
even within the @xmath1 expansion , there are no results available on the phase structure of the 2 + 1 dimensional gn model beyond the leading order , due to the appearance of complicate integrals ( note however that there are results up to @xmath9 for the critical exponents @xcite of the bulk transition ) .
it then becomes an important and interesting issue to investigate whether there is indeed a tricritical point for the gn model at finite @xmath2 in 2 + 1 dimensions and by which method we would be able to localize it in a robust form . by doing so , we expect to have a much better understanding of the chiral symmetry restoration problem at finite @xmath10 and @xmath11 not only in this model but also to have eventually a better knowledge of a similar behavior that may happen in qcd or even in analog condensed matter systems . in this work
we study the 2 + 1 dimensional gn model beyond the large-@xmath2 limit and , to obtain its phase diagram at finite @xmath10 and @xmath11 , we make use of the method of the optimized perturbation theory ( opt ) ( also known as linear @xmath12 expansion ) @xcite . here , we discuss how this method can be used to extract useful information , regarding critical quantities , from the landau expansion of the free energy density .
this paper is organized as follows . in the next section
we introduce the model , presenting the opt free energy density . in sec .
iii , we discuss optimization issues related to the landau expansion of the free energy density . our results for the @xmath13 phase diagram as well as for the critical quantities are presented in sec .
iv , showing the existence of tricritical point for finite values of @xmath2 . our conclusions and perspectives for future applications
are discussed in sec .
the gn model is described by a lagrangian density for a @xmath2 component fermion field @xmath14 ( @xmath15 ) with local quartic fermion interaction which is defined by @xcite ( in this work all our expressions are defined in euclidean space ) @xmath16 where , in 2 + 1 dimensions , the fermion fields are treated as four - component dirac spinors and the @xmath17 matrices are defined in the @xmath18 representation . in this way , eq .
( [ gn ] ) is invariant under the discrete chiral transformations @xmath19 and @xmath20 . though it also has a continuous flavor symmetry we will only be interested in the case of discrete symmetry , recalling a general no - go theorem in 2 + 1 dimensions that forbids spontaneous symmetry breaking of a continuous symmetry at any finite temperature @xcite .
the implementation of the opt within this model is reviewed in a previous application @xcite to the 1 + 1 dimensional case . in practice
, one considers the original theory , eq .
( [ gn ] ) , adding a quadratic ( in the fermion field ) term @xmath21 where @xmath22 is an arbitrary mass parameter and @xmath12 is a bookkeeping parameter which labels the order at which the opt is being performed . at the same time , @xmath23 . also , as conventional
, we rewrite the quartic term by introducing an auxiliary scalar field @xmath24 .
then , eq . ( [ gn ] ) acquires its opt form @xcite @xmath25 the opt then describes an interpolated theory with a fermion propagator @xmath26^{-1}$ ] , yukawa vertex @xmath12 and @xmath24 propagator @xmath27 .
any quantity @xmath28 is then evaluated to an order-@xmath29 using these new feynman rules . as shown in the many previous applications using this method @xcite
, it operates for any @xmath2 and is free from infrared divergences .
nonperturbative results for the original ( @xmath22-independent ) theory are obtained by optimizing @xmath28 with respect to @xmath22 at @xmath30 .
one standard and largely adopted optimization procedure is the variational criterion known as the principle of minimal sensitivity ( pms ) which fixes @xmath22 by requiring , order by order , that @xmath31 @xcite . from eq .
( [ gndelta ] ) , we are interested in evaluating the effective potential for the auxiliary scalar background field , @xmath32 , obtained once all fermion fields and fluctuations around the scalar background field are integrated out @xcite . from this quantity
all the thermodynamics for the model can be derived . in particular ,
csb is signaled by a non - vanishing vacuum expectation value for @xmath24 , @xmath33 , which is a minimum of @xmath34 . in fig .
[ diagrams ] we show all the feynman diagrams contributing to @xmath35 up to @xmath36 .
note that already at this first order in the opt a correction that goes beyond the large-@xmath2 expansion is included .
the first two diagrams in fig .
[ diagrams ] are @xmath37 , while the third one contains an exchange type of self - energy as insertion and brings the first @xmath38 correction to @xmath39 .
explicitly , we have for all the terms shown in fig .
[ diagrams ] the expression @xmath40 where the euclidean momenta are given by @xmath41 and @xmath42 , with @xmath43 . in the above equation
we have used the compact notation : @xmath44 with all momentum space integrals done with dimensional regularization , @xmath45 , while renormalization can be implemented in the modified minimal subtraction scheme ( @xmath46 ) .
all the contributions considered in eq .
( [ veff ] ) are actually finite in dimensional regularization . in ref .
@xcite we have considered the free energy up to order-@xmath47 , at @xmath48 and @xmath49 , where the first divergences show up and we refer the interested reader to that work for a detailed account on the renormalization problem of the 2 + 1 dimensional gn model in the opt .
let us only emphasize here that though the model is not perturbatively renormalizable , to @xmath50 and within dimensional regularization , we only need standard ( i.e. mass , wave - function , etc ) renormalization counterterms to render the effective potential calculation finite @xcite .
the effective potential , @xmath35 , is a natural quantity to be optimized in the spirit of the pms procedure since it represents the system s ( landau ) free energy density , whose knowledge allows for a complete analysis of all relevant thermodynamical properties .
also , as explained in ref .
@xcite , by optimizing the free energy we can guarantee that not only the exact large-@xmath2 results are already obtained at leading order in the opt , but also at any subsequent order , provided that one stays within the @xmath51 limit ( at first order in the opt this corresponds to dropping the third diagram in fig .
[ diagrams ] and applying the optimization procedure to the first two diagrams ) .
this shows that the opt is convergent at any order in the opt when contrasted with the large-@xmath2 predictions .
this result is valid for any temperature or value of chemical potential and applies as well to any number of dimensions ( for other accounts on the convergence of the opt in the context of quantum field theory , please see @xcite and references therein ) .
let us consider the expression for @xmath39 at @xmath36 with @xmath52 in which case chiral symmetry is broken at @xmath53 . after taking the traces in eq .
( [ veff ] ) , rearranging the terms and employing the redefinition of the coupling , @xmath54 , one obtains @xmath55 ^ 2 - \delta \frac{2 \pi}{n \lambda } [ { \cal i}_2(\eta,\mu , t ) ] ^2 \,\ , , \label{vlde1ddim}\ ] ] where @xmath56 setting @xmath30 and applying the pms to eq .
( [ vlde1ddim ] ) we obtain a self - consistent equation for the optimized mass @xmath57^{-1 } \right\ } \bigr|_{\eta = { \bar \eta } } \,\ , , \label{pmsselfconsistent}\end{aligned}\ ] ] where the primes indicate derivatives with respect to @xmath22 .
it can easily be verified that one immediate consequence of eq .
( [ pmsselfconsistent ] ) is that for @xmath58 we have @xmath59 , thus retrieving all large-@xmath2 results , as mentioned in the previous paragraph .
all the details concerning the matsubara sums and dimensional regularized integrals can be found in ref .
( note that the sum over the matsubara frequencies and all space momentum integrals can be evaluated analytically at this first @xmath12 order here considered ) . in the case
@xmath49 , eq . ( [ pmsselfconsistent ] )
factorizes in a nice way which allows us to understand how the opt resums the series producing non - perturbative results . using @xmath60 ( see ref .
@xcite ) we can write eq .
( [ pmsselfconsistent ] ) in the self - consistent form :
@xmath61 explicitly , @xmath62 is given by ( from the matsubara sum and momentum integration ) @xmath63 + \ln \left[1 + e^{-(|{\bar \eta}| + |\mu|)/t}\right ] \right\ } \;.
\label{intp2}\ ] ] the resummation becomes even easier to be visualized if one notes that @xmath64 , where @xmath65 represents an exchange ( rainbow , fock ) type of contribution so that , at @xmath49 , one may write @xmath66 .
note that prior to optimization , the terms contributing to @xmath67 to first order are the ones given by eq .
( [ veff ] ) , represented by the diagrams shown in fig .
[ diagrams ] . upon optimization
the diagrams are dressed by rainbow type of self - energy contributions that originates from the first nontrivial diagram represented by the third term in fig .
[ diagrams ] .
this was expected since at order-@xmath12 the perturbative opt effective potential receives information about this type of topology only .
the opt will then gather all powers of @xmath1 corresponding to the ( rainbow ) class of diagram .
in other words , the simple evaluation of a first topologically distinct diagram is already able to bring non - perturbative information concerning that type of contribution .
if one proceeds to order-@xmath47 , information about corrections to the scalar propagator as well as yukawa vertex will enter the perturbative effective potential .
then , the pms will dress up these perturbative contributions and so on .
of course this rather simple opt / pms procedure does not give all the truly non - perturbative and higher order contributions , such as the genuine @xmath68 and higher order corrections , but is just one particular class of graphs that are being resummed in this way .
it nevertheless gives a non trivial result that goes beyond the standard @xmath1 order .
let us now consider the @xmath69 case . by optimizing @xmath35 , at @xmath36
, it is very easy to show that one finds the optimum value @xmath70 , where @xmath71 is the minimum of @xmath35 , and @xmath72 is a characteristic function that will appear in all our expressions . at this value of @xmath22 and computing the extremum for the effective potential for @xmath73
, we obtain that the order parameter is given by @xmath74 note that the scalar field vacuum expectation value @xmath71 sets the natural scale for csb .
we could express all our other results in terms of @xmath71 , but here we find more convenient to normalize all critical quantities in terms of just @xmath75 ( after all this is just a scaling choice ) . in ref .
@xcite the result for the critical temperature has been obtained in a similar and completely analytic fashion producing the opt result @xmath76\;. \label{tc}\ ] ] however , an interesting alternative is to obtain @xmath3 using the landau expansion for the free energy density , which is valid for small values of the order parameter .
the landau s expansion ( for small @xmath73 ) for @xmath77 can be expressed in the general form @xmath78 where @xmath79 is a constant ( field independent ) energy term .
note that only even powers of @xmath73 are allowed due to the original chiral symmetry of the model .
the coefficients @xmath80 and @xmath81 appearing in eq .
( [ landau ] ) can be obtained , respectively , by a second , fourth and sixth derivative of the free energy density expansion at @xmath82 .
higher order terms in the expansion ( [ landau ] ) can be verified to be much smaller than the first order terms and can be consistently neglected .
lets recall that the expression of the free energy in the form ( [ landau ] ) is the simplest one able to exhibit a rich phase structure , allowing also for studying tricritical phenomena .
in particular note that a tricritical point can emerge whenever we have three phases coexisting simultaneously .
for the above potential , eq .
( [ landau ] ) , one can have a second order phase transition when the coefficient of the quadratic term vanishes ( @xmath83 ) and @xmath84 .
a first order transition happens for the case of @xmath85 .
the tricritical point occurs when both the quadratic and quartic coefficients vanish , @xmath86 ( with @xmath87 ) .
thus , eq . ( [ landau ] ) offers a simple and immediate way for analyzing the phase structure of our model .
for instance , to obtain @xmath3 at @xmath49 one only needs to consider eq .
( [ landau ] ) to order-@xmath88 with @xmath89 to assure that the potential is bounded from below .
then , the solution of @xmath90 sets the critical temperature .
however , in order to use landau s expansion we must have @xmath67 in terms of @xmath73 , @xmath11 and @xmath10 only ( apart from @xmath2 and the scale @xmath75 , of course ) . in principle , this can be done by using the pms relation , eq .
( [ pmsselfconsistent ] ) .
as one can easily check the large @xmath2 result , @xmath91 is quickly recovered by solving @xmath90 with the latter obtained from eq .
( [ vlde1ddim ] ) with @xmath92 .
now , at finite @xmath2 , @xmath93 depends on @xmath73 in a highly nonlinear way and the use of landau s method in conjunction with the opt - pms does not appear to be straightforward . however , eq . ( [ pmseta ] )
can then be easily solved numerically by iteration .
let us consider the first order ( in the iteration ) approximate pms solution ( pms1 ) @xmath94 and to access the convergence of the solutions , let us also consider the next approximation ( pms2 ) by doing @xmath95 on the right hand side of eq .
( [ pmseta ] ) . in table
i we compare the results given by pms1 and pms2 with the analytical relation @xmath96 $ ] , eq .
( [ tc ] ) , for @xmath49 .
.the critical temperature , @xmath3 ( at @xmath49 and in units of @xmath75 ) , obtained with the opt in three alternative ways for different values of @xmath2 .
we have considered as much digits as we needed to establish eventual numerical differences . [
cols="^,^,^,^ " , ] it is also possible to express and analyze the phase diagram in terms of the pressure @xmath97 and density @xmath98 , defined respectively by @xmath99 and @xmath100 .
this was done in @xcite , where it was shown the existence of a chiral broken / restored coexistence phase ( the analog of a liquid - gas phase ) , whose result is a direct consequence of the existence of the tricritical point seen in fig . [ phase ] . at this point one
may wonder how the consideration of the order-@xmath47 terms , which contains additional @xmath1 as well as new @xmath68 corrections , would affect the first order results .
clearly , at @xmath101 and @xmath102 the explicit evaluation of all the three loop contributions is a rather difficult task .
nevertheless , at @xmath48 and @xmath49 , a lengthy evaluation of those contributions has been done in ref .
@xcite showing that the potential corrections to the @xmath36 results are not negligible but remain under control , about a few percent .
for example , the result obtained for the scalar field vacuum expectation value at @xmath48 and @xmath49 is found to be ( at @xmath103 ) @xmath104 at @xmath105 , for a reasonable choice of the arbitrary renormalization scale , chosen as given by @xmath75 , while at @xmath106 it is @xmath107 . for larger values of @xmath2
the difference gets even smaller , in conformity with the large-@xmath2 results ( since at @xmath108 the opt converges exactly to the large-@xmath2 results @xcite ) .
this then hints on the accuracy and quick convergence of the order-@xmath12 results discussed in this letter .
in conclusion , by studying the 2 + 1 dimensional four - fermion gn model with discrete chiral symmetry in terms of the opt , we have shown how this method can be advantageous compared to the widely used large-@xmath2 and monte carlo approximations .
here we have presented precise analytical expressions for the order parameter @xmath71 ( at @xmath69 ) , the critical temperature @xmath3 ( at @xmath49 ) , and the critical chemical potential @xmath109 ( at @xmath48 ) for csr which are valid for any values of @xmath2 already at lowest order in the opt . as usual
, all these relations exactly reproduce the standard large-@xmath2 results when the @xmath58 limit is taken .
we were also able to precisely locate a tricritical point in the @xmath13 phase diagram not seen using the large-@xmath2 approximation and also not obtained before using other methods , though numerical simulations seemed to indicate its presence .
we have discovered that , due to the finite @xmath2 effects , the model behaves as a van der waals liquid in which a coexistent liquid - gas phase does exist .
the observed metastable region is rather small , which possibly explains why its observation was missed in ref .
@xcite and previous works .
all these results seem to survive when going to higher orders in the opt , as we discussed in the last section . in this letter
we have extended our previous applications of the opt to the gross - neveu model @xcite by analyzing how the method can be applied to the landau expansion for the free energy density in order to determine the value of critical quantities in a way that is less time consuming than the straight numerical methods used in refs .
we have shown that this is possible if one considers further approximations in the optimization equations .
however , the numerical effects of such approximations , even at the first level , turn out to be negligible . concerning the next order opt corrections we have checked in ref .
@xcite that , for @xmath53 , @xmath103 , and a reasonable choice of renormalization scale the differences for the order parameter are about @xmath110 . as for the @xmath1 expansion we are not in position to speculate what a complete next to the leading @xmath68 order calculation would give from a quantitatively perspective .
however , we believe that it should point out to the existence of a tricritical point in the @xmath111 plane since , according to our work , this seems to be an effect of finite @xmath2 corrections which is also supported by monte carlo applications @xcite
. possible follow - ups to this work that would be interesting to investigate regard the possible existence of a kind of nuclear matter state around @xmath109 , at large @xmath11 and low @xmath10 , for which the chiral phase would be degenerate with a baryonic phase , @xmath112 and @xmath113 .
this is still an unsettled question in the literature and that could be ideally addressed with the opt .
another interesting work would be to evaluate the various critical ( and tricritical ) exponents , determining whether they belong to the mean - field universality class or the ising universality class in two dimensions .
this is also a controversial point concerning the gn model , with no clear answer so far in the literature .
based on its recent success in dealing with critical theories @xcite we expect that the method employed here will be a valuable tool in these analysis providing us with a better understanding of the chiral symmetry behavior in a hot and dense medium .
finally , we point out that since the gn model is part of a larger class of four - fermi models , e.g. the nambu - jona - lasinio type of models commonly used in nuclear physics problems and as an effective model for qcd , or even its non - relativistic version , used in the study of planar superconductors and low dimensional fermionic systems ( like fermi gases in atomic physics ) , we expect that not only our results but also the opt method used to obtain them can be of interest and have applications in these areas as well .
in particular , we believe that the success of the opt method in redrawing the phase diagram of the three dimensional gn model could certainly attract the attention of condensed matter as well as nuclear physicists to this method as an alternative tool to the well known large-@xmath2 and monte carlo techniques , which are currently used in an interdisciplinary manner .
s. christofi and c. strouthos , arxiv : hep - lat/0612031 ; s. hands , j. b. kogut , c. g. strouthos and t. n. tran , phys . rev .
* d68 * ( 2003 ) 016005 ( 2003 ) ; s. j. hands , a. koci and j. b. kogut , nucl .
b390 * ( 1993 ) 355 .
yukalov , teor . mat .
* 28 * ( 1976 ) 92 ; r. seznec and j. zinn - justin , j. math .
phys . * 20 * ( 1979 ) 1398 ; a. okopiska , phys
* d35 * ( 1987 ) 1835 ; a. duncan and m. moshe , phys . lett . * b215 * ( 1988 ) 352 .
m. b. pinto and r. o. ramos , phys . rev .
* d60 * ( 1999 ) 105005 ; _ ibid . _
* d61 * ( 2000 ) 125016 ; j .- l .
kneur , a. neveu and m. b. pinto , phys . rev . *
a69 * ( 2004 ) 053624 ; s. k. gandhi and m. b. pinto , phys . rev . * d49 * ( 1994 ) 4258 ; s. k. gandhi , h. f. jones and m. b. pinto , nucl . phys . *
b359 * ( 1991 ) 429 ; c. arvanitis , f. geniet , m. iacomi , j .- l . kneur and a. neveu ,
a12 * ( 1997 ) 3307 ; k. g. klimenko , z. phys .
* c50 * ( 1991 ) 477 . | the method of optimized perturbation theory ( opt ) is used to study the phase diagram of the massless gross - neveu model in @xmath0 dimensions . in the temperature and chemical potential plane ,
our results give strong support to the existence of a tricritical point and line of first order phase transition , previously only suspected to exist from extensive lattice monte carlo simulations .
in addition of presenting these results we discuss how the opt can be implemented in conjunction with the landau expansion in order to determine all the relevant critical quantities .
keywords : chiral phase transition , gross - neveu model , four - fermi models , nonperturbative thermodynamics |
in networks , packets have to be routed between nodes through a series of intermediate relay nodes . each intermediate node in the network may receive packets via multiple data streams that are routed simultaneously from their source nodes to their respective destinations . in such conditions , packets may have to be stored at intermediate nodes for transmission at a later time .
if buffers are unlimited , intermediate nodes need not have to reject or drop arriving packets .
however , in practice , buffers are limited in size .
although a large buffer size is preferred to minimize packet drops , large buffers have an adverse effect on the _ latency _ , i.e. , the delay experienced by packets stored in the network .
further , using larger buffer sizes at intermediate nodes would also result in secondary practical issues such as increased memory - access latency .
though our work is motivated by such concerns , our work is far from modeling realistic conditions .
this work modestly aims at providing a theoretical framework to understand the fundamental limits of single information flow in finite - buffer line networks and investigates the tradeoffs between throughput , packet delay and buffer size .
the problem of computing capacity and designing efficient coding schemes for lossy wired and wireless networks has been widely studied @xcite . however , the study of capacity of networks with finite buffer sizes has been limited .
this can be attributed solely to the fact that analysis of finite buffer systems are generally more challenging . with the advent of network coding as an elegant and effective tool for attaining optimum network performance ,
the interest in finite - buffer networks has increased @xcite .
the problem of studying lossy networks with finite buffers has been investigated in the area of queueing theory under a different but similar framework .
the queueing theory framework attempts to model packets in a network as customers , the delay due to packet loss over links as service times in the nodes , and the buffer size at intermediate nodes as the queue size .
further , the phenomenon of packet overflow in communications network is modeled by blocking ( commonly known as _ type ii _ or _ blocking after service _ ) in queueing networks @xcite
. however , this packet - customer equivalence fails in general network topologies due to the following reason .
when the communications network contains multiple disjoint paths from the source to the destination , the source node can choose to duplicate packets on multiple paths to minimize delay .
this replicating strategy can not be captured directly in the customer - server based queueing model .
therefore , the queueing framework can not be directly applied to study packet traffic in general communications networks . however , queueing theory offers solid foundation for studying buffer occupancies and packet flow traffic in line networks .
there has been extensive study in queueing theory literature on the behavior of open tandem queues , which are analogous to line networks @xcite .
however , approaches from queueing theory literature predominantly consider a continuous - time model for arrival and departure of customers / packets . in this work
, we consider a discrete - time model for packet arrival and departure processes by lumping time into epochs .
this model is similar to those in @xcite .
the broad contributions of this paper can be summarized as follows .
the bulk of this work operates under the assumption of perfect hop - by - hop feedback .
we present a markov - chain based model for exact analysis of line networks .
the capacity of a line network is shown to be related to the steady - state distribution of an underlying chain , whose state space grows exponentially in the number of hops in the network .
simple assumptions of renewalness of intermediate packet processes are employed to estimate the capacity of such networks .
the estimates are exact for two - hop networks . however , the estimates extend the results of @xcite to networks of any number of hops and buffer sizes of intermediate nodes . using the estimates ,
the profile of packet delay is derived and studied .
using the exact markov chain model in conjunction with network coding , it is shown that the throughput capacity is not affected by the absence of feedback in line networks .
this result is similar to the information - theoretic result that feedback does not increase capacity of point - to - point channels @xcite .
finally , simulations reveal that our estimates closely predict the trends and tradeoffs between hop - length , buffer size , latency , and throughput in these networks .
this paper is organized as follows .
first , we present the formal definition of the problem and the network model in section [ fb - sec1 ] .
next , we present our framework for analyzing capacity of finite - buffer line networks in section [ fb - sec2 ] . the proposed markovian framework is then employed to investigate packet delay in section [ fb - delay ] .
we compare our analytical results with simulations in section [ fb - sec4 ] and conclude it with a brief discussion on the inter - dependence of buffer usage , capacity and delay .
finally , section [ conc ] concludes the paper .
this work focuses on the class of line networks . as illustrated in fig .
[ fig1 ] , @xmath0 denotes the number of hops in the network , and @xmath1 and @xmath2 to denote the set of nodes and the set of links in the network , respectively .
such a network has @xmath3 intermediate nodes , which are shown by black squares in the figure .
each intermediate node @xmath4 is assumed to have a buffer of @xmath5 packets .
note that buffer sizes of different nodes can be different . without loss of generality , we assume @xmath6 and @xmath7 , for @xmath8 . further , it is assumed that the destination node has no buffer constraints and that the source node possesses an infinitude of innovative packets at all times .
the system is analyzed using a discrete - time model , where each node can transmit at most one packet over a link in any epoch .
intermediate buffers are assumed to be empty at epoch @xmath9 and the dynamics for @xmath10 are steered by the loss processes on the edges of the network .
the loss process on each link is assumed to be memoryless and statistically independent of the loss processes on other links .
we let @xmath11 to denote the erasure probability on the link @xmath12 for @xmath13 . in this model
, a node receives a packet on an incoming link when the neighboring upstream node transmits a packet and when the packet is not erased over the link .
the reader is directed to appendix [ app.0-discvscts ] for a discussion on how the assumed discrete - time model relates to continuous - time exponential model that is commonly employed in queueing theory . for the bulk of this work , we assume that the network has a perfect hop - by - hop feedback mechanism indicating the transmitting node of the receipt and storage of the transmitted packet by the receiving node
. however , a subsequent section of this paper drops this assumption to study the capacity of line networks without feedback .
it is also assumed in this work that nodes operate in a _ transmit - first _
mode , i.e. , each node first generates a packet ( if it has a non - empty buffer ) and transmits it on the outgoing edge .
the node then processes the buffer after receiving the acknowledgement from the next - hop node before accepting / storing the packet on its incoming edge . for notational convenience ,
the random process on the link @xmath14 is denoted by @xmath15 .
@xmath16 if and only if the packet transmitted at epoch @xmath17 is deleted by the channel @xmath18 , and @xmath19 otherwise . for the sake of succinctness , we let @xmath20 and buffer sizes @xmath21 . the focus of this paper is two - fold .
the foremost aim is to identify the _ supremum _ of all rates that are achievable by the use of any coding strategy between the ends of a line network with erasure probabilities @xmath22 and buffer sizes @xmath23 . in the line network illustrated in fig .
[ fig1 ] , we first aim to identify the maximum rate of information that the node @xmath24 can transmit to node @xmath25 , which is denoted by @xmath26 . the next issue on which we focus is the delay experienced by packets in intermediate node buffers when the network operates near the throughput capacity . in our analysis , we employ the following notations .
vectors will be denoted by boldface letters , eg . , @xmath27 .
the indicator function for the set @xmath28 is represented by @xmath29 $ ] . for any @xmath30 $ ] , @xmath31 .
the convolution operator is denoted by @xmath32 and @xmath33 is used as a shorthand for the @xmath17-fold convolution of @xmath34 with itself . for @xmath35 , @xmath36 denotes the probability mass function of a positive random variable that is geometric with mean @xmath37 . for a discrete random variable @xmath38 with probability mass function @xmath39 , @xmath40 and @xmath41
are both used to denote the mean of the random variable @xmath38 .
lastly , for appropriate @xmath42 , @xmath43 denotes the galois field of size @xmath44 .
in this section , we investigate the effect of finite buffers on the capacity of line networks .
first , we present a framework for exact computation of the capacity of line networks that have perfect hop - by - hop feedback . we then present bounds on the capacity using techniques from queueing theory .
subsequently , we present our approaches to approximate the capacity of a line network . in the concluding subsection , we illustrate that the throughput capacity remains unaltered when feedback is absent provided packet - level network coding is allowed .
the problem of identifying capacity is related to the problem of identifying schemes that are _ rate - optimal_. in the presence of lossless hop - by - hop feedback , the scheme performing the following steps in the given order is rate - optimal . *
if the buffer of a node is not empty at an epoch , then it must transmit one of the stored packets at that time . *
a node deletes the packet transmitted at an epoch if it receives an acknowledgement of packet storage from the next - hop node at that epoch . * after performing @xmath45 and @xmath46 , a node accepts an arriving packet if it has space in its buffer and sends an acknowledgment of packet storage to the previous node .
notice that in the above scheme , at each epoch , the buffer of the last intermediate node is updated first , and the buffer of the first intermediate node is updated last . to determine the throughput capacity of the network , we need to track the number of packets that each node possesses at every instant of time by using the rules of buffer update under the above optimal scheme .
let @xmath47 be the vector whose @xmath48 component denotes the number of packets @xmath4 possesses at time @xmath17 .
the variation of state at the @xmath49 epoch can be tracked using auxiliary random variables @xmath50 defined by @xmath51x_i(l ) & i = h\\ x_i(l)\sigma[n_{i-1}(l)(m_{i}-n_{i}(l)+y_{i+1}(l ) ) ] & 1<i
< h\\ x_i(l)\sigma[m_{i}-n_{i}(l)+y_{i+1}(l ) ] & i=1 \end{array}\right.\hspace{-2mm}.\label{fb - eqn2}\end{aligned}\ ] ] from the definition of the auxiliary binary random variables in ( [ fb - eqn2 ] ) , we see that @xmath52 only if all the following three conditions are met : * node @xmath53 has a packet to transmit to @xmath54 .
* the link @xmath18 does not erase the packet at the @xmath49 epoch , i.e. , @xmath55 , and * node @xmath54 is not full after its buffer update due to its transmission over @xmath56 at the @xmath49 epoch .
the changes in the buffer states can then by seen to be given by the following .
@xmath57 note that since @xmath58 is a function of @xmath59 and @xmath60 , @xmath61 depends only on its previous state @xmath59 and the channel conditions @xmath62 at the @xmath49 epoch .
hence , @xmath63 forms a markov chain .
the number of states corresponds to the number of possible assignments to @xmath64 , which amounts to @xmath65 possibilities .
however , since at each time instant the number of packets that can be transmitted over any link is bounded by unity , we see that for every @xmath66 , @xmath67 therefore , the number of non - zero entries in each row of the probability transition matrix term of the matrix @xmath68 represents the probability that the next state is @xmath69 given that state is presently @xmath70 . ]
@xmath68 representing the transitions in the occupancy is bounded above by @xmath71 . a detailed categorization of the states @xmath72 that enables further understanding can be performed thus .
we can order the states of the chain in such a way that the state @xmath73 corresponds to the row index @xmath74 in the matrix @xmath68 .
denote @xmath75 to be the set of states that have @xmath76 for @xmath77 .
let @xmath78 represent the transition matrices for transitions from states in @xmath75 to those in @xmath79 , @xmath75 , @xmath80 , respectively .
then , it can be shown that @xmath81 , @xmath82 , and @xmath83 for @xmath84 ( see lemma [ fb - lem1 ] ) .
therefore , the transition matrix of the chain can be structurally represented as follows .
@xmath85 the dynamics given by the above equation can be depicted pictorially by the chain in fig .
note that due to the finite buffer condition and the non - negativity of occupancy , the transitions from the first block and from the last block differ from the transitions from the blocks between them .
further , the states within each @xmath86 , @xmath87 , can be organized into @xmath88 sets in a similar fashion .
in addition to this structural property , the transition sub - matrices satisfy the following algebraic properties .
[ fb - lem1 ] in a generic line network , the following hold .
* @xmath89 , @xmath90 , and @xmath91 for @xmath92 . * for @xmath6 ,
@xmath93 is non - singular and upper triangular for @xmath94 . * for @xmath95 , @xmath96 is singular and lower triangular for @xmath97 .
* @xmath98 is non - singular @xmath99 .
see appendix [ app.1 - 1 ] .
to illustrate the implications of the above lemma , consider the markov chain for a three - hop line network with erasure probabilities @xmath100 , and with buffer sizes @xmath101 presented in fig .
[ fb - mc ] . for this network ,
the algebraic properties of lemma [ fb - lem1 ] can be understood as follows . *
any transition involving a decrease in the second component involves a non - negative change in the magnitude of the first component . *
any horizontal transition involving a decrease in the second component is always feasible ( provided the second component of the starting state is positive ) . *
any transition involving an increase in the second component involves a non - positive change in the magnitude of the first component . *
not all horizontal transitions involving an increase in the second component are feasible .
for example , the transitions from the state @xmath102 to @xmath103 and from the state @xmath103 to @xmath104 are infeasible , and hence @xmath105 .
while the first two facts relate to the upper triangular structure and non - singularity of @xmath106 , the latter two relate to the lower triangular and singularity properties of @xmath107 .
this markov chain for the dynamics of the state of a line network with perfect feedback is _ irreducible , aperiodic , positive - recurrent _ , and _ ergodic _ @xcite . by ergodicity
, we can obtain temporal averages by statistical averages .
therefore , the throughput capacity @xmath26 can be identified by appropriately scaling the likelihood of the event that the system is in a state wherein the last node buffer is non - empty .
this quantity is given by @xmath108 \label{fb - eqn5.1 } \end{aligned}\ ] ] notice that packets are not erased from the buffers without a receipt of acknowledgement of storage from the next - hop node .
therefore , the packet - flow rate is conserved .
therefore , for @xmath109 , the throughput capacity can also be identified from @xmath110 .
\label{fb - eqn5.1b } \end{aligned}\ ] ] thus , the problem of identifying the capacity of line networks is reduced to the problem of computing the steady - state probabilities of the aforementioned markov chain . however , due to the size of the markov chain and its transition matrix , and the presence of multiple reflections due to the limited buffers at intermediate nodes , the problem of computing the steady - state distribution and capacity is computationally tedious even for networks of reasonable hop - lengths and buffer sizes . as the first step towards estimation ,
we can define a finite sequence of matrices by @xmath111 note that these matrices relate the steady - state distribution @xmath112 of the states in @xmath86 , @xmath87 by @xmath113 . using these relations ,
we can , in theory , estimate the capacity by @xmath114 however , this matrix - norm approach does not provide insight into occupancy statistics of various nodes .
therefore , we focus on an approximations - based approach to capacity estimation in the remainder of this work . in queueing theory , problems of identifying the steady - state probability of stochastic networks have often been dealt with approximations .
most approaches to problems in this area have been to approximate the dynamics of the network by focussing on local dynamics of the network around each node and the edges incident with it .
the key idea in this section is to modify the exact markov chain to derive bounds on throughput capacity .
to do so , notice that the main reason for intractability of the exact system is the strong dependence of @xmath50 on not only @xmath115 , but also @xmath116 .
this dependence translates to a strong dependence of @xmath117 on both @xmath118 and @xmath119 .
relaxation of this strong dependence will be a step towards possible decoupling of the system , and a deeper understanding of the tradeoffs in such networks .
consider a network operation mode where each intermediate note transmits an acknowledgement whenever it _ receives _ a packet ( as opposed to the rate - optimal setting where it sends an acknowledgement whenever it _ receives and stores _ a packet successfully ) . under this new mode of operation
, we notice that the dependence of the state of @xmath48 node on that of nodes further downstream is eliminated .
this mode of operation is equivalent to assuming that a packet that arrives at a node whose buffer is full gets lost / dropped unlike the optimal mode of operation where it gets re - serviced . in this mode ,
the state updates are given by a simplified markov chain that is generated by the following rule for all @xmath120 and @xmath121 .
@xmath122\hspace{-0.45mm}-\hspace{-0.45mm}\tilde{y}_{i+1}(l ) , \hspace{-1.5mm}\label{amceqn}\end{aligned}\ ] ] where @xmath123x_i(l ) & 1<i\leq h\\ x_i(l ) & i=1\end{array}\right ..
\label{fb - eqn6}\end{aligned}\ ] ] to avoid confusion , we appellate the chain that is obtained by the dynamics defined by ( [ fb - eqn2 ] ) and ( [ fb - eqn3 ] ) as the exact markov chain ( emc ) and the one defined by ( [ amceqn ] ) and ( [ fb - eqn6 ] ) as the approximate markov chain ( amc ) . also , we allow @xmath59 and @xmath124 to always denote the state of an instance of the process generated by the emc and the amc , respectively
. then , the following property holds . _
( temporal boundedness property of the amc)_[fb - thm1 ] consider a line network with @xmath0 hops and an instance of channel realizations @xmath125 .
suppose we track the variation of the states of the emc and the amc using this instance of channel realizations with the same initial state @xmath126 .
then , for any @xmath120 and @xmath127 , the following holds .
@xmath128 the proof is detailed in appendix [ app.1 - 2 ] .
the temporal boundedness property guarantees that statistically , the probability that a node has an empty buffer is overestimated by the amc .
in fact , if we can identify the steady - state distribution of the states of amc , we can provide a lower bound for the steady - state probability of any subset of states @xmath129 that have the form @xmath130 where @xmath131 for @xmath132 .
using the temporal boundedness property in conjunction with ( [ fb - eqn5.1 ] ) , we can provide a lower bound @xmath133 for the capacity of the line network by underestimating the probability in ( [ fb - eqn5.1 ] ) by using the steady - state distribution of the amc instead of that of the emc .
equivalently , the capacity of the line network is at least that of the throughput achievable by the amc .
this above idea of lower bound extends easily to an upper bound using the following result .
the fundamental idea behind the following bound is to manipulate the buffer sizes at each node so that the packet drop in the modified network is provably infrequent than in the actual network .
[ fb - ubthm ] let the operator @xmath134 be defined by @xmath135
. for a given network with distinct erasure probabilities @xmath22 and buffer sizes @xmath23 , denote @xmath136 to be the throughput computed from the steady - state distribution of the amc defined by ( [ amceqn ] ) and ( [ fb - eqn6 ] ) with erasure probabilities @xmath22 and buffer sizes @xmath137 , i.e. , @xmath138 . then , @xmath139 a detailed proof is presented in appendix [ app.1 - 4 ] .
thus , the problem of bounding capacity is reduced to identifying the steady - state probability of the amc .
notice that the above bounds are not in a computable form , since they still involve identifying the steady - state distribution of the amc .
even though the amc is significantly simpler than the emc , the output process from each intermediate node is not renewal @xcite .
therefore , the distribution of inter - departure times from each intermediate node is insufficient to completely describe the arrival process at intermediate nodes @xmath4 for @xmath140 .
therefore , a straightforward hop - by - hop analysis ( without further assumptions ) seems insufficient to identify the capacity of such networks . in this section , we present two iterative estimates for the capacity of line networks that is based on certain simplifying assumptions regarding the emc .
we notice that the difficulty of exactly identifying the steady - state probabilities of the emc stems from the finite buffer condition that is assumed .
the finite buffer condition introduces a strong dependency of state update at a node on the state of the node that is downstream .
this effect is caused by blocking when the state of a node is forced to remain unchanged because the packet that it transmitted is successfully delivered to the next - hop node , but the latter is unable to store the packet due to lack of space in its buffer .
additionally , the non - tractability of the emc is compounded by a non - renewal packet departure process from each intermediate node . in this section ,
we ignore some of these issues to develop iterative methods for estimation .
figure [ fb - ie ] encapsulates the assumptions made in both estimation approaches . while both approaches ignore the non - renewal nature of packet arrival process at each node , the first approach makes an additional memoryless assumption on the arrival process .
additionally , both approaches model the effect of blocking by the introduction of a single parameter @xmath141 that represents the probability that an arriving innovative packet will be blocked .
this estimate makes the following assumptions to decouple the dynamics of the system and enable capacity estimation . *
the packet departure process at each intermediate node is memoryless . in other words ,
each node @xmath4 sees a packet arrival process that is memoryless with ( average ) rate @xmath142 packets / epoch .
this assumption allows us to track information rates over links while simplifying the higher order statistics . *
any packet that is transmitted unerased by the channel @xmath12 is blocked independently with a probability @xmath143 .
that is , for any @xmath144 , @xmath145 = { \overline}{\varepsilon}_{i+1}{p_b}_{i+1}.\nonumber\end{aligned}\ ] ] + here , @xmath143 denotes the blocking probability due to full buffer state at @xmath146 .
this assumption allows us to track the blocking probability ignoring higher order statistics of the blocking process . * for each node @xmath4 and epoch @xmath17 , the event of packet arrival and the event of blocking from @xmath146 are independent of each other . under the simplifying assumptions.,width=326 ] the above assumptions are valid in the limiting case of large buffers provided the system corresponds to a stable queueing configuration . by assuming that they hold in general , the effect of blocking is spread equally over all non - zero states of occupancy at each node .
similarly , the assumptions also spread the arrival rate equally among all occupancy states .
given that the arrival rate of packets at the node @xmath4 is @xmath142 packets / epoch , and the blocking probability of the next node is @xmath143 , the local dynamics of the state change for the node @xmath4 under assumptions a1-a3 is given by the markov chain of fig . [ fb-1hmc ] with the parameters set to the following .
@xmath147 using these parameters , the steady - state distribution , then we set @xmath148 . ]
@xmath149 of the chain of fig .
[ fb-1hmc ] can be computed to be @xmath150 assuming that @xmath4 observes a packet arrival rate of @xmath142 from @xmath53 and a blocking probability of @xmath143 from @xmath146 , the blocking probability @xmath151 that the node @xmath53 perceives from the node @xmath4 and the arrival rate @xmath152 that @xmath146 observes can be computed via ( [ fb - ss ] ) using the following equations .
@xmath153 note that the blocking probability @xmath151 is computed using the full occupancy probability of the node @xmath4 .
while in reality , a packet is blocked by @xmath4 only if at the arriving instant , the node has full occupancy , a2 models any arriving packet to be blocked with the above probability irrespective of the occupancy of @xmath4 .
also , in ( [ fb - pb ] ) and ( [ fb - r ] ) the arrival rate from the node @xmath154 is @xmath155 and the blocking probability @xmath156 . given two vectors @xmath157^{h}$ ] and @xmath158^{h}$ ] ,
we term @xmath159 as a rate - approximate solution to emc , if they satisfy the equations ( [ fb - pb ] ) and ( [ fb - r ] ) in addition to having @xmath155 and @xmath156
. since these relations were obtained from making assumptions on the emc , it is _ a priori _ unclear if there exist rate - approximate solutions for a given system @xmath160 .
fortunately , the following result guarantees both the uniqueness and an algorithm for identifying the rate - approximate solution to the emc .
[ fb - uniqthm ] given a line network with link erasures @xmath161 and intermediate node buffer sizes @xmath162 , there is exactly one rate - approximate solution @xmath163 to the emc .
further , the rate - approximate solution satisfies flow conservation .
that is , @xmath164 the proof is detailed in appendix [ app.1 - 3 ] finally , the estimate of the capacity can be obtained from the rate - approximate solution by computing the average rate of packet storage at each node using @xmath165 note that by the conservation of flow , any @xmath166 can be used in the above equation to identify capacity . as an illustration ,
consider a simple four - hop network with erasures @xmath167 and buffer sizes @xmath168 . from the above estimation method
, we arrive at @xmath169 from simulations , the throughput capacity was found to be @xmath170 packets / epoch for the same network . in this section , we assume that the given line network @xmath171 satisfies @xmath172 for @xmath173 . since the capacity of a line network is a continuous function of the system parameters , this assumption is not restrictive .
a system with non - distinct erasure parameters can be approximated to any degree of precision by a system with distinct erasure probabilities .
before we introduce the second approach for estimation , we present the following technical results or all @xmath174s be positive . we only need that their sum be unity and that they generate a valid probability distribution , respectively . ] wherein we denote @xmath175 to be the identity distribution for the convolution operator .
[ fb - lbthm ] consider a tandem queueing system of two nodes where the first node possessing @xmath176 buffer slots is fed by a renewal process whose inter - arrival time distribution is @xmath177 with @xmath178 and @xmath179 for @xmath180 .
suppose that the distribution of service time is @xmath181 , where @xmath182 for @xmath183 .
further , suppose that the second node blocks an arriving packet memorylessly with probability @xmath184 , and that any blocked packet gets re - serviced .
then , the distribution of inter - arrival times as seen by the second node is given by @xmath185 where @xmath186 for some @xmath187 and @xmath188 , @xmath189 , with @xmath190 .
a detailed analysis including the means of identifying @xmath191 is given in appendix [ app.2 ] .
just as in the rate - based iterative estimate , this estimate also makes three assumptions to simplify the emc . while the distribution - based iterative
estimate makes assumptions a2 and a3 , it relaxes assumption a1 to the following : * the packet departure process at each intermediate node is renewal .
note that assumption a1 allows for tracking only the average rate of information flow on edges whereas a1@xmath192 allows tracking of the distribution of packet inter - arrival times .
however , a1@xmath192 ignores the fact that the distribution of an inter - arrival time changes with the knowledge of past inter - arrival times .
to track the inter - arrival distribution and blocking probabilities at each node , the distribution - based iterative estimate uses theorem [ fb - lbthm ] in a hop - by - hop fashion .
assuming that the packet arrival process at @xmath4 is renewal with an inter - arrival distribution @xmath193 , and that the memoryless blocking from @xmath146 occurs with probability @xmath143 , we see that the packet inter - arrival distribution seen by @xmath146 is given by @xmath194 notice that just like in ( [ eqn-24 ] ) , @xmath195 uses the effective erasure probability to incorporate the effect of blocking by @xmath146 .
however , this corrective term does not appear in @xmath196 term , because @xmath197 represents the distribution of packet inter - arrival times at @xmath146 , and not the distribution of the time between two adjacent successful packet storages at @xmath146 .
further , the blocking probability of @xmath4 as perceived by @xmath53 is given by @xmath198\\ & \stackrel{(\ref{fb - blockingeqn})}{= } \mathcal{p}(f_i , m_i,{\varepsilon}_{i+1},{p_b}_{i+1 } ) .
\label{fb - pb - die}\end{aligned}\ ] ] just as in the rate - based iterative estimate , we call a solution to ( [ fb - r - die ] ) and ( [ fb - pb - die ] ) with boundary conditions @xmath156 and @xmath199 as a distribution - approximate solution .
though the existence and uniqueness of the distribution - approximate solution for a given system @xmath171 has eluded us , simulations reveal that for each system , the solution is unique and can be found by iteratively using the following algorithm .
note that during any round of @xmath200 in the above algorithm , ( [ fb - r - die ] ) can be iteratively used to identify @xmath201 $ ] in step 5 only if the output distribution of inter - departure times from each node is a weighted sum of geometric distributions .
this is however guaranteed if the erasure probabilities of no two links are equal .
alternately , step 2 can be replaced by a convergence - type criterion instead of the ` max_iter ` criterion .
after sufficiently large number of iterations , the distributions and blocking probabilities usually converge ( to @xmath202 and @xmath203 ) , and upon convergence the capacity can be estimated via @xmath204 using the above approach for the four - hop example network at the end of sec .
[ fb - sec2.3.1 ] , we have @xmath205 feedback from next - hop node provides a natural means of buffer update and packet deletion . in the absence of feedback , due to the finiteness of buffers ,
each intermediate node must have a local rule for buffer update to accept packets that arrive .
a rule for packet deletion or update must be maintained at each node so that the buffers are used efficiently .
a network coded - scheme based on random linear combinations over a large finite field @xmath206 of size @xmath44 as is described in @xcite presents an effective means of buffer update and packet delivery .
consider the following scheme based on network coding .
* at each epoch , a node having a buffer size of @xmath176 packets picks a vector @xmath207 uniformly at random to generate a random linear combination in the following manner . for each buffer slot @xmath208 ,
the packet @xmath209 stored in that slot is represented as a vector over @xmath206 and the output packet is generated by computing @xmath210 .
this generated packet is then transmitted during the epoch . *
if a packet @xmath211 is received by a node at an epoch , it first generates the output packet at that instant and then updates its buffer in the following manner .
the node selects a vector @xmath212 uniformly at random and for each @xmath213 , adds the packet @xmath214 to the packet stored in the @xmath215 buffer slot .
note that in the network coding scheme described above , after sufficient time after the commencement of packet transfer from the source , all buffer slots of every intermediate node almost always have non - trivial contents unlike the scheme with perfect feedback .
however , it is not true that all of these packets are _ innovative _ , i.e. , packets may contain common _ _ information _ _ is said to contain @xmath216 packets of information if @xmath217 . ] .
such a condition may occur when the packets are linearly dependent in the algebraic sense . with this notion of information
, the rate of information received by the destination node can be seen to be the asymptotic rate of arrival of innovative packets .
the following result characterizes these rates achieved by the network coding scheme over the field @xmath206 and relates it to throughput capacity in the presence of lossless feedback @xmath26 .
[ thm - nfopt ] let @xmath218 denote the rate of arrival of innovative packets at the destination node of a line network without feedback assuming that the aforementioned network coding scheme over the field @xmath43 is employed . then , for each sequence of finite fields @xmath219 such that @xmath220 , we have @xmath221 the proof is presented in appendix [ app.3 ] . from ( [ fb - eqn1 ] ) , we observe that there is no loss in achievable rates when feedback is absent and that the aforementioned network coding scheme is rate - optimal for line networks without feedback , provided a large field size is employed .
in this section , we use the iterative estimates of section [ fb - sec2.3 ] to obtain estimates on the probability distribution of the delay experienced by information packets in line networks with perfect feedback under the optimal strategy of section [ fb - sec2.1 ] .
we abstain from defining latency of data packets in networks without feedback , since optimal schemes for such networks involve packet - level coding .
when perfect feedback is available , we define the delay of a packet as the time taken from the instant when the packet is stored in the buffer of the first intermediate node to the instant when the destination receives it .
since the delay statistics depend on how the packets are handled in intermediate nodes , in addition to the optimal scheme of section [ fb - sec2.1 ] , we assume a _ first - come first - serve _ treatment of packets at intermediate node buffers .
note that this assumption is made only for the ease of presentation .
the framework permits the analysis of randomized schemes where each node after a successful transmission selects a packet in its buffer randomly and memorylessly , and transmits it repeatedly until it is stored at the next - hop node . in order to compute the distribution of delay that a packet experiences in the network
, one can proceed in a hop - by - hop fashion using two parameters : ( 1 ) @xmath222 , an estimate of blocking probability at node @xmath4 , @xmath223 , and ( 2 ) @xmath224 , an estimate of the distribution of occupancy at node @xmath4 ( for each @xmath8 and @xmath225 ) just before packet arrival conditioned on the event that the arriving packet is successfully stored . in the last relay node @xmath226 , the additional delay perceived by a packet arriving at @xmath227 epoch depends on the occupancy @xmath228 of the node @xmath226 and @xmath229 .
suppose at epoch @xmath17 , node @xmath226 has @xmath230 packets excluding the arriving packet .
then , the packet has to wait for the @xmath231 already - stored packets to leave before it can be serviced .
since the services are memoryless , the distribution of delay is given by a sum of @xmath232 independent geometric distributions each with a mean inter - arrival time @xmath233 , i.e. , @xmath234 .
hence , the distribution of additional delay induced by waiting in the buffer of @xmath226 is @xmath235 .
\label{last_relay_delay}}\ ] ] however , the situation is different for other intermediate delays because of the effect of blocking .
the additional delay incurred while being stored in the node @xmath236 , is given by @xmath237 } , \label{int_relay_delay}\ ] ] since a packet is deleted from the buffer of @xmath238 only if the channel successfully transmits it and @xmath239 does not block the arriving packet , which by assumption a2 occurs memorylessly with a probability @xmath240 . assuming that the delays incurred by waiting in the buffer of each node is independent of each other
, we obtain the total delay considering all hops to be @xmath241 note that in addition to the above delay , the source node attempts to transmit the packet multiple times before the packet is successfully accepted at the first intermediate node .
the distribution of this time spent in this is approximately given by a random variable whose distribution is @xmath242 .
thus , the iterative estimation technique provides us with a framework to approximately but analytically compute the delay profile using an estimate of distribution of packets seen by an arriving packet that is successfully stored , and an estimate of the blocking probabilities .
finally , the pair of estimates ( @xmath243 , @xmath222 ) can be obtained from the rate - approximate solution by using @xmath244 similarly , another pair of estimates can be obtained using the distribution - based iterative estimate by @xmath245 where @xmath246 is the eigenvector of the ( [ app2-eqn1 ] ) upon convergence . combining the above equations , two estimates for the delay profile for line networks with feedback
can be obtained .
in this section we present the results of simulation comparing our analytic results to simulations of line networks with perfect feedback .
first , the simulations for the capacity are presented , and then the simulations for delay profiles are presented .
this section ends with a discussion on the efficient usage of buffers and the interplay of buffer size , capacity and delay . in our model ,
a line network is completely defined by the number of hops , the erasure probability for each link and the buffer size at each intermediate node . to study the accuracy of our bounds and estimates ,
we vary one of these three parameters while keeping the remaining two fixed . in each of the figures , the actual capacity and bounds obtained via simulations
are presented in addition to our estimates .
further , for the sake of brevity , we abbreviate distribution - based iterative estimate ( algorithm [ alg : dbie ] ) ) , rate - based iterative estimates ( algorithm [ alg : rbie ] ) , lower bound ( thm .
[ fb - thm1 ] ) , and upper bound ( thm .
[ fb - ubthm ] ) to dbie , rbie , lb , and ub , respectively .
figure [ fb - linel ] presents the variation of the capacity and varying hop - length @xmath0.,width=326,height=216 ] with the number of hops of line networks when each intermediate node possesses a buffer size of five packets .
the figure presents simulations for networks when the probability of erasure on each link is set to either @xmath247 or @xmath248 .
first , it is noticed that the bounds and iterative estimates agree with the actual capacity for two - hop networks .
second , it is noticed that the bounds and estimates capture the variation of the actual capacity of the network .
however , the estimates are more accurate . for both choices of channel parameters ,
both estimates predict throughput capacity within an error of 1% .
further , it is also noticed from the figures that the independence assumptions of the estimates generally over - estimate the actual capacity of the network . in order to study the effect of buffer size on capacity
, we simulated a five - hop line network with each link having erasure probabilities just as in the previous setting .
figure [ fb - linem ] presents the variation of our results and the with varying buffer size @xmath176.,width=326 ] actual capacity as the buffer size of the intermediate node is varied .
it can be seen that as the buffer size is increased , all curves approach the ideal min - cut capacity of @xmath249 . also , as is expected , the accuracy of the bounds improve with the buffer size . finally , the effect of the channel conditions on the capacity of a five - hop line network with intermediate buffer sizes of five packets each is presented in figure [ fb - linee ] .
it is noticed that as the , @xmath250 with varying erasure probability @xmath251.,width=326 ] probability of erasure increases , the loss in capacity due to finite buffer becomes more pronounced .
for example , for the simulation setting of fig .
[ fb - linee ] , the loss in capacity varies from @xmath252 at @xmath253 to @xmath254 at @xmath255 in a near - linear fashion . from these figures
, we infer that it is paramount that the effect of blocking be considered as realistically as possible . modeling the effect of blocking as packet loss ( as is done to derive our bounds ) only allows us to loosely bound the capacity of such networks .
figure [ delaydist - line ] presents the variation of delay profile for the optimal strategy in an eight - hop line network with the erasure probability on every link set to @xmath247 .
the delay profiles were simulated for three different buffer sizes . as in section
[ fb - delay ] , the estimate and simulations were performed for the first - come first - serve strategy . from the first sub - plot , it is noticed that both mean and variance of the delay distribution increase as buffer sizes increase . while the mean delay obtained via simulations for the three memory settings
are @xmath256 , @xmath257 , and @xmath258 epochs , whereas the analytical result for the same using the dist .- based iterative estimate are @xmath259 , @xmath260 , and @xmath261 epochs , respectively .
note that the analytical estimates for the mean delay @xmath262 can be obtained without computing the delay profile by the use of little s theorem @xcite as follows .
@xmath263 where , as before , @xmath264 denotes the distribution of occupancy of @xmath4 at steady state given by the rate - based estimate .
note that each term in the above sum can be viewed as the contribution of the corresponding node to overall delay .
it is noted that the analytic prediction of the delay profile is more conservative than the actual delay profile in the sense that the estimate of the variance is higher than the actual variance of packet delay .
the second sub - plot of the figure illustrates the difference in the cumulative distribution of delay predicted by the two estimates .
it is noticed from all the above simulations that there is only a minor difference between the two estimation schemes if the parameters of interest are either the throughput capacity or the delay profile .
figure [ fb - pdf ] highlights the difference between the two estimates when continuous - time models are emulated using discrete - time epochs . consider a three - hop line network where intermediate nodes have a buffer of three packets and their packet service distributions are exponential with @xmath265 .
suppose that the arrival process at the first node is renewal with inter - arrival distribution being exponential with @xmath266 .
the following figure presents the distribution of inter - departure duration from the second node .
it is observed that by lumping @xmath267 seconds into each epoch , the distribution - based iterative estimate provides a near - accurate distribution of the inter - departure durations . on the other hand ,
the rate - based iterative estimate approximates the distribution as an exponential , which yields a less accurate estimate .
note that for this setting @xmath268 packets / sec , and the distribution - based and rate - based estimates are 2.2447 and 2.2413 packets / sec , respectively . in this section
, we present a brief discussion on two questions pertaining to efficient usage of buffers in intermediate nodes .
_ is the use of more buffer slots , the merrier ? _ and _ how to allocate buffers to different nodes so that operation ensures near - min - cut throughput and acceptable delay ?
_ to address the first question , consider the eight - hop network of fig .
[ delaydist - line ] . as the buffer size is varied from 10 to 15 packets , the rate - based estimate for capacity changes from 0.7135 to 0.7254 packets / epoch
a change of less than 1.5% ( of the min - cut bound )
. however , the mean latency changes from 55.18 to 81.29 epochs a 47% change .
therefore , for each @xmath22 , it is likely that there is a critical buffer size for each node beyond which the throughput capacity improvement is marginal ; however , with increase in buffer sizes , the average time packets spend in the network continues to grows significantly .
one must therefore identify the correct size of buffers to be used so that both latency and throughput capacity are acceptable . to discuss the second issue
, we illustrate with the following example .
consider a four - hop network with @xmath269 $ ] for which a good choice of buffer allocation needs to be identified under the constraint that the total number of buffers in the network must be no more than 30 packets . to this end
, we use the rate - based estimate to study the effect of individual buffer sizes on throughput and delay .
[ delay - throughput ] shows the variation of the throughput and delay contributed by each node when its memory is varied from 1 to 20 packets , while the buffer sizes of other intermediate nodes are kept at 20 packets . in this example
, it is noticed that maximum throughput estimate for all choices of memory estimates is @xmath270 packets / epoch when @xmath271 .
this setting offers a mean packet delay of @xmath272 epochs .
however , minimum delay configuration amongst those that offer a throughput more than @xmath273 packets / epoch is @xmath274 , which offers a throughput of @xmath275 packets / epoch and a mean packet latency of 28.46 epochs . the actual capacity and delay for these configurations were found to be @xmath276 packets / epoch , @xmath277 epochs and @xmath278 packets / epoch , @xmath279 epochs , respectively . to understand further these patterns , we present in fig .
[ buffer - occupancy ] the steady - state occupancy of the three intermediate nodes when buffer sizes are set to @xmath280 packets , @xmath281 packets and @xmath282 packets , respectively . in all settings
, it is noted that the node @xmath154 is congested because the sub - network from @xmath154 to @xmath283 has a min - cut capacity of @xmath248 , whereas it receives packets at the rate of @xmath284 .
therefore , the steady - state occupancy of the node @xmath154 for @xmath285 and @xmath286 are translates of that of @xmath287 . due to congestion ,
an arriving packet at such a node usually sees very high occupancy .
hence , in a first - come first - serve mode of operation , the arriving packet has to wait long before getting serviced .
therefore , it is critical that the buffer size of congested nodes ( such as @xmath154 ) be kept to absolute minimum to minimize average packet delay .
similarly , @xmath288 can at most receive packets at a rate of @xmath248 , however the outgoing link can communicate packets at a much higher rate .
therefore , the buffer of @xmath288 is never full as long as the buffer size is greater than five .
nodes such as @xmath288 that are never congested contribute little to the delay experienced by packets .
hence , limiting buffer sizes of such nodes is not critical for delay as long as the sizes are bigger than their threshold sizes ( beyond which throughput increase is marginal ) .
occupancy in nodes like @xmath289 that are neither congested nor starved undergo non - trivial changes with changes in buffer sizes .
these nodes contribute significantly to both the throughput and average packet delay in the network .
for example , in the example network @xmath289 has a near - uniform distribution for both @xmath285 and @xmath287 . just like congested nodes ,
such nodes have to be allocated buffer sizes so that the they neither block packets nor contribute to delay significantly .
though the classification of nodes as congested , starved or neither can usually be done by focusing on @xmath22 , good memory allocation requires knowledge of trends of latency and throughput with buffer sizes , which in turn require the help of more sophisticated estimates such as those proposed in this work . as a second example , consider another four - hop network with @xmath290 $ ] . in the infinite buffer setting ,
the queueing system corresponding to this buffer configuration is stable .
hence , no node can be classified _ a priori _ as congested .
suppose that a throughput - optimal allocation of buffer sizes for intermediate nodes is to be designed with the constraint that the total number of packets in the network be limited to 60 . clearly , a nave first guess is to assign @xmath291 $ ] .
however , notice that no matter how large the buffer sizes are , the probability of blocking at any node is always non - zero .
hence , the rate of arrival that @xmath289 and @xmath288 see is smaller than that noticed by @xmath154 .
therefore , it is meaningful to assign @xmath154 a larger buffer size to minimize blocking at @xmath154 and maximize throughput .
although this intuition is correct , it is unclear as to how to allocate buffers .
the strength of the iterative technique is in resolving exactly this issue by assigning estimates to each buffer allocation configuration . by searching around the neighborhood of @xmath292 ,
the maximum throughput configuration is found to be @xmath293 $ ] .
as is illustrated by these examples , the proposed iterative estimation techniques presents a framework to identify nodes in line networks that are either : ( a ) starved and therefore play an insignificant role in capacity and packet delay ( such as @xmath288 of fig .
[ buffer - occupancy ] ) , or ( b ) congested and contribute significantly to packet delay ( such as @xmath154 of fig .
[ buffer - occupancy ] ) , or ( c ) contribute significantly to both capacity and packet delay ( such as @xmath289 of fig .
[ buffer - occupancy ] ) . on identifying these nodes ,
it is possible to identify configurations that make efficient use of the buffers without severely compromising on either throughput capacity or average packet delay .
this work focused on the effect of finite buffers on the throughput capacity and packet delay profile in line networks with packet erasure links .
first , an exact markovian framework for modeling line networks with perfect feedback was presented .
the framework was simplified using independence assumptions to derive iterative estimation techniques that yield approximations of all marginal buffer statistics and also allow to identify the packet delay profile in such networks .
further , it was shown that the absence of feedback has no effect on the throughput capacity of line networks provided packet - level coding is permitted . finally , via simulations ,
the proposed iterative techniques were noticed to be computationally - efficient and near - accurate models to analyze and study the behavior of line networks .
in this section , we argue that the discrete model assumed in the paper can be used to study the capacity of tandem queue model with type ii blocking ( see @xcite ) and independent exponential service times at each node . consider a tandem queue of @xmath0 links and @xmath3 intermediate nodes .
suppose @xmath294 denotes the parameters for the exponential service times at @xmath295 , respectively .
assuming that each intermediate node has buffers given by @xmath296 and that @xmath297 denotes the number of packets collected by @xmath25 in the period @xmath298 , the throughput capacity @xmath299 ) hinges on the ergodicity of the continuous - time system .
] of the system is defined by @xmath300 can be computed from a discrete model assumed in this paper .
for example , fig .
[ ctsvsdisc - throughput ] considers a four - hop system with each node having exponential processing times with parameter @xmath301 s@xmath302 and compares it with four discretized models .
note that the approximations become finer as smaller values of @xmath303 are chosen .
this fact can be formalized as follows .
@xmath304 where the right - hand side uses the discrete - time model of ( [ fb - eqn5.1 ] ) .
we begin by constructing the probability transition matrices for continuous and discrete chains that track the state of the system just before a departure from the last intermediate node .
note that both chains use the same state space @xmath305 , however their transition probabilities are different .
let @xmath306 denote the probability transition matrix for the continuous model .
let @xmath307 denote the transition matrix that effects the change in states when a departure from the last intermediate node occurs and let @xmath308 denote the transition matrix corresponding to changes in state over a duration of @xmath309 seconds given that no departure occurs in that duration .
then , @xmath310 where @xmath311 denotes the cumulative density function of the exponential rv with parameter @xmath312 .
notice that for any @xmath313 , @xmath314
. therefore , @xmath315 where ( a ) follows since @xmath316 .
let @xmath317 denote the state transition matrix for one time epoch of the discretized model with @xmath318 .
then , we have @xmath319 therefore , we have @xmath320\bigg),\end{aligned}\ ] ] where ( a ) follows from fubini - tonelli theorem @xcite .
the discrete equivalent @xmath321 of the above transition matrix that tracks the state between departures for the corresponding time - discretized system is given by @xmath322\end{aligned}\ ] ] let @xmath323 for any pair of states @xmath324 , @xmath325 .
then , @xmath326\nonumber\\ & \stackrel{(c)}{\leq}\lim_{\tau\rightarrow 0 } \bigg[\sum_{i=0}^\infty ( \lambda_{h}\tau)\big|e^{-i\lambda_{h}\tau}-({\overline}{\lambda_{h}\tau})^{i}\big| + \frac{e(\lambda_{h}\tau)^2}{{\overline}{e^{-\lambda_{h}\tau}}}\bigg]\nonumber\\ & = \lim_{\tau\rightarrow 0 } \bigg[\sum_{i=0}^\infty ( \lambda_{h}\tau)\big|e^{-i\lambda_{h}\tau}-(1-\lambda_{h}\tau)^{i}\big|\bigg]\nonumber\\ & \stackrel{(d)}{=}\lim_{\tau\rightarrow 0 } \bigg[(\lambda_{h}\tau)\sum_{i=0}^\infty \big(e^{-i\lambda_{h}\tau}-(1-\lambda_{h}\tau)^{i}\big)\bigg]\nonumber\\ & = \lim_{\tau\rightarrow 0 } \big(\frac{\lambda_{h}\tau}{1-e^{-\lambda_{h}\tau}}-1\big)=0.\end{aligned}\ ] ] note that ( b ) follows since @xmath327 is a probability matrix and hence each component is bounded above by unity , and ( c ) follows since for @xmath328 , it is true that @xmath329 and ( d ) follows from @xmath330 for @xmath331 .
thus @xmath332 as @xmath333 .
let @xmath334 and @xmath335 be the eigenvectors of @xmath336 and @xmath306 , respectively .
then , since the steady - state distribution of a chain is a continuous function of the transition matrix , it follows that @xmath337 , as @xmath333 .
however , the capacity computed using continuous and discrete models are given by @xmath338 therefore , @xmath339 .
\(a ) suppose that the state of the system is @xmath340 with @xmath341 , then from ( [ fb - eqn2 ] ) , we notice that @xmath342 and @xmath343x_{h-1}$ ] .
hence , given the event @xmath344 , @xmath345 depends only on @xmath346 and @xmath347 and not on @xmath228 .
this guarantees that @xmath348 for @xmath349 .
\(b ) first suppose @xmath95 .
consider @xmath93 for some @xmath350 and the state of the system at some time @xmath351 .
@xmath93 represents transitions from states that have the form @xmath352 to states of the form @xmath353 . since @xmath354
, it must be that @xmath355 and that the channel must have erased the packet transmitted by @xmath356 .
denote @xmath357 for @xmath358 and @xmath359 .
then , it is seen that for any realization of @xmath360 , it is true that the state transition must obey @xmath361 however , @xmath362 is the index of the row corresponding to the state @xmath59 within @xmath93 and @xmath363 is the index of the column corresponding to @xmath61 within @xmath93 .
therefore , all possible transitions in @xmath93 correspond to transitions from states to other state that involve a non - positive change in the row - index .
therefore , @xmath93 is upper triangular .
finally , since each diagonal term of @xmath93 is bounded below by @xmath364 , we conclude that @xmath365 finally , if @xmath366 , it is easy to see that @xmath367 $ ] .
\(c ) consider a transition under @xmath96 for @xmath368 from a state that has the form @xmath352 to another that has the form @xmath369 after an epoch . since @xmath370
, it must be that the packet transmitted during this epoch on the link @xmath371 must have reached successfully , i.e. , @xmath372 . by an argument similar to the above one , we can show that @xmath96 is lower triangular .
however , certain diagonal terms are zero . in specific , consider the transition from state @xmath373 to the state @xmath374 which corresponds to the @xmath375 .
however , this transition is impossible when @xmath95 , since the node @xmath356 has no packets to send during this epoch .
thus , @xmath376 if @xmath377 .
\(d ) the non - singularity of @xmath98 follows from the fact that @xmath378 is diagonal dominant @xcite , since @xmath379 . on the other hand , since @xmath380 , there exists at least one @xmath231 for which the inequality is strict , which guarantees the non - singularity of these matrices .
we proceed by mathematical induction on the time index @xmath17 .
clearly , the condition holds for @xmath9 .
suppose that the claim is true for all nodes and for times @xmath381 for some @xmath382 .
consider the states of the node @xmath4 for some @xmath383 in both chains at time instant @xmath231 .
one of the two following cases must apply .
* @xmath384 : in this case , we note that @xmath385 + if @xmath386 , then @xmath387 and @xmath388-\sigma[{{\tilde{n}}}_{i-1}(k)]\geq 0.\nonumber\end{aligned}\ ] ] thus @xmath389 .
+ now , if @xmath390 , it is seen from ( [ fb - eqn2 ] ) and ( [ fb - eqn6 ] ) that @xmath391 . further , if @xmath392 , then clearly , @xmath393 and @xmath394 . + if @xmath395 , then @xmath396 and @xmath397 follows since @xmath398-\sigma[{{\tilde{n}}}_{i-1}(k)]\big)\geq 0.\nonumber\end{aligned}\ ] ] + now , if @xmath399 , then ( [ fb - eqn2 ] ) and ( [ fb - eqn6 ] ) again imply @xmath391 and @xmath400-\sigma[{{\tilde{n}}}_{i-1}(k)]\big)\geq 0 $ ] , and hence @xmath401 follows .
* @xmath402 : assume let @xmath403 .
then , @xmath404x_i(k)\nonumber\\ & \leq{{\tilde{n}}}_i(k)+ 1 - y_{i+1}(k ) \leq n_i(k)-y_{i+1}(k)\nonumber\\ & \leq n_i(k)+y_i(k)-y_{i+1}(k ) = n_i(k+1).\nonumber\end{aligned}\ ] ]
lastly , if @xmath405 , then the claim can be violated only if @xmath406 and @xmath407 , which can happen only if @xmath408 . however , under this channel instance , @xmath409 .
thus , @xmath410 .
thus , we have the following .
@xmath411 the proof is then complete by following the above argument for @xmath154 and interpreting @xmath412=\sigma[{{\tilde{n}}}_{0}(k)]=1 $ ] , since the source always possesses innovative packets .
if @xmath366 comparing ( [ fb - eqn2 ] ) and ( [ fb - eqn6 ] ) , we see that the amc and the emc are identical .
hence , we may assume @xmath95 .
the proof in this case is based on mathematical induction on the time index @xmath17 . at each time
, we compare the state of the emc with that of the modified amc .
let the extended state of the emc at an instant @xmath413 be denoted by @xmath414 , where the notation is identical to that of sec .
[ fb - sec2 ] with the addition that @xmath415 denotes the number of packets that the destination has received by the @xmath49 epoch .
similarly define the extended state of the amc with modified buffer sizes at an instant @xmath413 by @xmath416 .
define a partial ordering of vectors of @xmath417 in the following manner . for two vectors @xmath418 ,
@xmath419 if @xmath420 for each @xmath223 .
we track the system starting from initial rest ( all buffers being empty ) using an instance of channel realizations .
clearly @xmath421 .
suppose that @xmath422 for @xmath423 .
consider @xmath424 .
one of the following two situations may arise , @xmath382 in this proof . ] .
* @xmath425 : in this case , no node is saturated in the amc and hence every node can potentially accept packets provided both the node preceding it has packets to send and the channel allows it . consider the number of packets that are in the buffers of nodes @xmath426 for some @xmath427 in both chains .
+ \(i ) if @xmath428 or if both @xmath429 and @xmath430 are true , then @xmath431 + \(ii ) if @xmath429 , @xmath432 and @xmath433 then @xmath434 therefore , @xmath435 + \(iii ) finally , if @xmath429 , @xmath432 and @xmath436 then @xmath437 since @xmath69 was arbitrary , it follows that @xmath438 . *
@xmath439 : then , let @xmath440 . in this case
, nodes @xmath441 are not saturated and can accept packets .
the argument for @xmath442 follows for @xmath443 is similar to the previous case .
notice that since the occupancy of nodes @xmath54 , @xmath444 are not full , @xmath445 now , for @xmath446 , two cases may occur .
+ \(i ) if @xmath447 , then by ( [ app1-comment ] ) , @xmath448 + \(i ) if @xmath449 , and @xmath450 then @xmath451 however , if @xmath449 and @xmath452 , then @xmath453 thus , the claim holds for @xmath454 .
the claim is then complete if @xmath455 .
therefore , in what follows , we may assume @xmath456 .
+ finally , for @xmath457 , one of the following cases must hold .
+ \(i ) if @xmath450 , then @xmath458 + \(ii ) if @xmath459 then @xmath460 and @xmath461q^e_1(k)\nonumber\\ & \geq \sum_{s > i } q^e_s(k-1)+\sum_{\iota=1}^i m_\iota+\sigma[2-j]\nonumber\\ & \geq\sum_{s > i }
n^e_s(k-1)+\sum_{s = j}^i n^e_s(k-1)+1\nonumber\\ & = \sum_{s\geq j } n^e_s(k-1)+1\geq \sum_{s\geq j } n^e_s(k).\end{aligned}\ ] ] thus , the claim is true for all indices @xmath462 and @xmath438 . here
, it must be noted that if the buffer sizes for the nodes of amc are not modified as in the hypothesis , ( [ app1-comment ] ) will not hold .
finally , the upper bound follows since @xmath463
consider two rate - approximate solutions @xmath464 and @xmath465 such that @xmath466 with @xmath467 .
notice that @xmath468 is a strictly decreasing function of @xmath469 when @xmath229 is kept fixed .
this follows from the fact that @xmath470 ^ 2 } < 0 .
\end{aligned}\ ] ] an easy way to understand this behavior is to notice that @xmath471 increase with @xmath469 , while @xmath472 decreases with @xmath469 .
therefore , from ( [ fb - r ] ) , it follows that @xmath473 now , from ( [ eqn - requiv ] ) and ( [ fb - pb ] ) guarantee @xmath474 .
we then use the monotonicity of @xmath475 in conjunction with already shown results to show that @xmath476 and @xmath477 . extending this inductively
, we have @xmath478 and @xmath479 .
therefore , for each @xmath480 , there is at most one solution satisfying @xmath481 . now , consider two rate - approximate solutions @xmath464 and @xmath465 such that @xmath482 . by monotonicity of @xmath468
, we have @xmath483 . from ( [ fb - pb ] ) , we notice that @xmath151 is also a strictly increasing function in both its variables @xmath142 and @xmath484 .
therefore , @xmath485 . again , proceeding inductively from the last node to the first each time noticing the monotonic growth of @xmath486 and @xmath487 , we conclude that @xmath488 however , since @xmath464 and @xmath465 are both rate - approximate solutions , we have @xmath489 , which contradicts ( [ ap1-contra ] ) .
therefore , there is at most one solution to the system of equations . to identify the unique solution
, we construct a sequence of tuples @xmath490,{\mathbf{p_b}}[l])\}_{l\in\mathbbm{n}}$ ] as described in algorithm [ alg : rbie ] . note that step 2 of the algorithm can be replaced by a convergence - type step that halts if @xmath491-\mathbf{r}[l-1]\|_1 $ ] is smaller than a chosen threshold . by the monotonic property of the non - linear system of equations ,
the following results can be established .
@xmath492 & < & r_i[l+1 ]
\\ { p_b}_i[l ] & < & { p_b}_i[l+1 ] \\
\end{array } , \quad\quad l\in\mathbbm{n}. \label{ap1-mont}\ ] ] however , each component of @xmath493 and @xmath494 is individually bounded by unity . therefore , the sequence of numbers for each component of these vectors must converge .
denote the component - wise limit as @xmath495 .
denote @xmath496^{h}\times[0,1]^{h}\longrightarrow [ 0,1]^{h}\times[0,1]^{h}$ ] to be the following map .
for each @xmath497^h$ ] , denote @xmath498 to be the pair , whose first component is the vector of rates computed from ( [ fb - r ] ) and the second component is the vector of blocking probabilities computed from ( [ fb - pb ] ) .
then , @xmath499 is a continuous map and @xmath500,{\mathbf{p_b}}[l]))=({\mathbf{r}}[l+1],{\mathbf{p_b}}[l+1])$ ] for each @xmath351 . also , for this sequence of rates and blocking probabilities , we note that @xmath501,{\mathbf{p_b}}[l])-\mathbf{w}^*\|_\infty \nonumber\\ & + \|\xi(({\mathbf{r}}[l],{\mathbf{p_b}}[l]))-({\mathbf{r}}[l],{\mathbf{p_b}}[l])\|_\infty \label{ap1-conv}\\ & + \|\xi(\mathbf{w}^*)-\xi(({\mathbf{r}}[l],{\mathbf{p_b}}[l]))\|_\infty .
\nonumber\end{aligned}\ ] ] however , the right - hand side of ( [ ap1-conv ] ) is true for any @xmath502 . by allowing @xmath503 ,
the three limits vanish and hence we see that @xmath495 is a fixed point of the map and hence the unique solution to the system of non - linear equations . finally , to see the conservation of flow , notice that the rate - based iterative estimate models the system using a discrete - time m / m/1/@xmath231 system by the introduction of additional assumptions and parameters . in the model ,
the number of innovative packets that are successfully stored by @xmath4 as the system progresses from @xmath9 to @xmath504 is given by @xmath505+{\overline}{\varepsilon}_{i+1}{\overline}{p_b}_{i+1}^*\pr[n_i = m_i]\big)+o(n)\nonumber\\ & = nr_i^*\big(1-\varphi(m_i|r_i^*,{\varepsilon}_{i+1},{p_b}_{i+1}^*)\big(1-{\overline}{\varepsilon}_{i+1}^*{\overline}{p_b}_{i+1}^*\big)\big)+o(n)\nonumber\\ & \hspace{-1.5mm}\stackrel{(\ref{fb - pb})}{= } nr_i^*{\overline}{p_b}_{i}^*+o(n)\label{fb - cons1}.\end{aligned}\ ] ] similarly , the number of packets successfully output by @xmath4 is given by @xmath506\big){\overline}{p_b}^*_{i+1}+o(n)\nonumber\\ & = n{\overline}{\varepsilon}_{i+1}^*\big(1-\varphi(0|r_i^*,{\varepsilon}_{i+1},{p_b}_{i+1}^*)\big){\overline}{p_b}^*_{i+1}+o(n)\nonumber\\ & \hspace{-1.5mm}\stackrel{(\ref{fb - r})}{=}nr_{i+1}^*{\overline}{p_b}^*_{i+1}+o(n)\label{fb - cons2}\end{aligned}\ ] ] since the m / m/1/@xmath231 system is lossless , all stored packets eventually leave the system .
thus , the average rate of packet storage at a node must match the average rate of packets output from that node . comparing ( [ fb - cons1 ] ) with ( [ fb - cons2 ] )
, the conservation of packet flow for the rate - approximate solution follows .
the proof elaborates the behavior of a tandem system via a formal setup for the discrete - time equivalent of the @xmath507 queue @xcite . to illustrate the complications in the setup , fig .
[ app2-fig1 ] presents a section of an inter - arrival period for the first node .
the number of customers in the queue of the node just before an arrival or a departure is presented on the axis .
the arrival and departure of customers is marked by incoming and outgoing arrows , respectively . in scenario
a , we see that the queue is never starved and as a result all the inter - departure times are instances of the service process .
however , in scenario b , we notice that all the five customers that are in the queue after the arrival are serviced much ahead of the next arrival and hence there is a period of time during which the queue is starved .
if the queue were not starved , it could have possibly serviced a customer at the instance marked by the outgoing dotted arrow .
hence , this duration of time denoted by @xmath508 in the figure , adds a delay to the inter - departure time .
thus , if we are able to extract the distribution @xmath509 of this duration , we can identify the inter - arrival distribution @xmath510 as seen by the second node to be a weighted sum of @xmath511 and @xmath512 . in order to identify the distribution @xmath513
, we need to identify the probability distribution @xmath514 of the number of customers in the first node s buffer just after an arrival .
the first step in identifying @xmath514 from the imbedded markov chain for the occupancy of the first node is to construct the distribution @xmath515 of the number of packets that could be potentially transmitted during an inter - arrival duration @xmath516 provided the queue were infinite .
this distribution can be computed from the arrival and departure processes in the following manner .
@xmath517\binom{k}{j}\tilde{{\zeta}}_{n}^{k - j}{\overline}{\tilde{\zeta}}_{n}^j \nonumber\\&=\sum_{k=1}^\infty \bigl(\sum_{l=1}^{n-1 } p_l { \overline}{{\zeta}}_l{\zeta}_l^{k-1}\bigr)\binom{k}{j}\tilde{\zeta}_{n}^{k - j}{\overline}{\tilde{\zeta}}_{n}^{j}\nonumber\\ & = \sum_{l=1}^{n-1 } p_l\frac{{\overline}{{\zeta}}_l}{{\zeta}_l}\big[\frac{{\overline}{\tilde{\zeta}_n}}{\tilde{\zeta}_n}\big]^j\bigl(\sum_{k=1}^{\infty } \binom{k}{j}({\zeta}_l\tilde{\zeta}_n)^k\bigr)\nonumber\\ & \stackrel{(a)}{=}\big[\frac{{\overline}{\tilde{\zeta}_n}}{\tilde{\zeta}_n}\big]^j\sum_{l=1}^{n-1 } \frac{p_l{\overline}{{\zeta}}_l}{{\zeta}_l}\big(\frac{({\zeta}_l\tilde{\zeta}_n)^j}{(1-{\zeta}_l\tilde{\zeta}_n)^{j+1}}-\sigma[1-j]\big),\end{aligned}\ ] ] where in the above , we use @xmath518 to incorporate the actual parameter of the memoryless service time , and in ( a ) , we use @xmath519 , @xmath520 . for each @xmath521 , the @xmath522 entry of the probability transition matrix @xmath523 for the imbedded markov chain that tracks the number of customers just after an arrival can be computed by @xmath524\bigl(\sum_{k = i}^\infty d_k\bigr)&+\sigma[j-1]d_{i+1-j}\nonumber\\&+\sigma[j - m+1]d_{i - j}.\label{app2-eqn1}\end{aligned}\ ] ] note that in ( [ app2-eqn1 ] ) , we set @xmath525 when @xmath526 .
the distribution @xmath514 can then be solved from the eigenvector relation @xmath527 .
note that a packet arriving at the first node will not be accepted if the node is in full buffer and no packet had left in the preceding inter - arrival duration .
the probability of this blocking event at the first node is given by @xmath528 finally , we can identify the distribution of @xmath508 by conditioning on the number of customers @xmath307 just after a customer arrival .
it is seen that for @xmath529 , @xmath530 & = \sum_{j=1}^\infty\hspace{-0.5 mm } \bigg[\hspace{-1.5mm}\begin{array}{ll}\pr[\textrm{the queue is emptied at time $ j$}]\\ \times \pr[t_a = i+j]\end{array}\hspace{-1.5mm}\bigg ] \nonumber\\ & = \sum_{j=1}^\infty\binom{j-1}{k-1}{\overline}{\tilde{\zeta}}_n^k{\tilde{\zeta}_n}^{j - k}\bigl[\sum_{l=1}^{n-1 } p_i { \overline}{{\zeta}}_l{\zeta}_l^{i+j-1}\bigr]\nonumber\\ & = \sum_{l=1}^{n-1}p_l{\overline}{\zeta}_l{\zeta}_l^{i-1}\bigl[\frac{{\overline}{\tilde{\zeta}}_n^k}{{\tilde{\zeta}_n}^k}\sum_{j=1}^\infty \binom{j-1}{k-1}({\zeta}_l{\tilde{\zeta}_n})^j\bigr]\nonumber\\ & = \sum_{l=1}^{n-1}\bigl(p_l\frac{({\zeta}_l{\overline}{\tilde{\zeta}}_n)^k}{(1-{\zeta}_l{\tilde{\zeta}_n})^k}\bigr)({\overline}{\zeta}_l{\zeta}_l^{i-1 } ) .
\label{app2-eqn3}\end{aligned}\ ] ] from ( [ app2-eqn3 ] ) , we notice that the distribution of @xmath508 conditioned on @xmath531 is a weighted sum of geometric distributions . the distribution of @xmath508 can then be computed as follows .
@xmath532}{\sum_{k=0}^m \pi_k \pr[x\geq 1|m = k ] } = \sum_{l=1}^l \beta_l { \overline}{\zeta}_l{\zeta}_l^{i-1},\nonumber\\ \beta_l&= p_l \bigg[{\mathop{\sum_{k\in\{0,\ldots , m\}}}_{l=\{1,\ldots , n-1\ } } } \hspace{-1 mm } p_l\frac{\pi_k({\zeta}_l{\overline}{\tilde{\zeta}_n})^k}{(1-{\zeta}_l{\tilde{\zeta}_n})^k}\bigg]^{-1}{{\sum_{0\leq k \leq m } } } \frac{\pi_k({\zeta}_l{\overline}{\tilde{\zeta}_n})^k}{(1-{\zeta}_l{\tilde{\zeta}_n})^k}\nonumber\end{aligned}\ ] ] also , we notice that the distribution of inter - arrival times @xmath533 as seen by the second node is either an instance of @xmath534 or that of @xmath512 , and hence can be written as @xmath535 for some @xmath536 $ ] .
the last step in constructing the inter - departure distribution is to identify @xmath537 .
this is done by noticing the mean duration between departures . over a large duration @xmath538 ,
the number of packets that are accepted at the first node is given by @xmath539 .
the number of packets that are accepted by the second node is given by @xmath540 .
since the system has finite buffer size and no loss , the rates must match .
therefore , one can identify @xmath537 using the following .
@xmath541 finally , notice that if @xmath542 , we have @xmath543 using the above we can see that @xmath544 which is also a weighted sum of geometric distributions .
we present below a fundamental result that will be used in various stages of the proof .
[ rlcinnolem ] let @xmath508 be a vector space over a finite field @xmath545 and let @xmath546 for any @xmath547 .
let @xmath548 and @xmath549 be two subsets such that @xmath550 .
then , let @xmath551 be selected uniformly at random from @xmath552 and set @xmath553 . then ,
@xmath554\hspace{-0.75mm}<\hspace{-0.75mm}\frac{q^{\dim({\mathscr{s}}(u)\cap{\mathscr{s}}(v))}}{q^{\dim({\mathscr{s}}(u))}}.\ ] ] let @xmath555 be the set of all vectors @xmath556 such that @xmath557
. then @xmath555 forms a commutative group under componentwise addition .
similarly , let for each @xmath558 , let @xmath559 be the set of vectors @xmath556 such that @xmath560 .
it is follows that @xmath561 , i.e. , they are the coset translates of the subgroup @xmath555 .
therefore , uniform selection of the coefficients to perform a linear combination results in the selection of a vector in @xmath562 uniformly at random .
notice that @xmath563 if and only if @xmath564 .
note that the occurrence of this event is improbable for large fields , since @xmath565=1-\frac{q^{\dim({\mathscr{s}}(u)\cap{\mathscr{s}}(v))}}{q^{\dim({\mathscr{s}}(u))}}.\nonumber\ ] ] the basic idea of the proof is to construct a chain for the setting without feedback that is similar to the emc .
once the chain is identified , the proof will be completed by showing that the transition probabilities of each transition approaches that of the emc as the field size is made large . to this end , allow @xmath571 to be the packet received by the node @xmath4 at the @xmath227 epoch . set @xmath572 if the @xmath14 channel erases the transmitted packet at the @xmath227 epoch . for the sake of proof ,
each epoch is divided into @xmath0 sub - epochs .
since the network is assumed to work in a transmit - first mode , the network updates the buffers in a reverse - hop fashion , i.e. , at the @xmath573 sub - epoch , the message generated by @xmath574 is used to update that of @xmath575 . define @xmath576 to be the set of packets in the buffer of @xmath4 after the @xmath573 sub - epoch of the @xmath49 epoch . for notational
ease , let @xmath577 for @xmath578 and @xmath579 .
note that the system is uniquely described by the dynamics of the nested vector spaces @xmath580 .
define occupancy for this coded setting as @xmath581 notice that this notion of occupancy denotes the number of additional _ innovative _ packets that is housed by @xmath4 at the @xmath227 epoch that has not been conveyed to downstream nodes . also note that at the @xmath573 sub - epoch of the @xmath49 epoch , the only buffer that changes is that of @xmath575 due to the receipt of @xmath582 .
thus , @xmath583 to investigate the change of occupancy ) and ( [ amceqn ] ) , occupancies @xmath584 and @xmath585 correspond to @xmath586 in this setup . ] after the @xmath573 sub - epoch of the @xmath49 epoch for nodes @xmath574 and @xmath575 , we have to consider the following cases . *
if @xmath587 , then by lemma [ rlcinnolem ] , we see that @xmath588 with probability at least @xmath589 . therefore , it follows that @xmath590 with high probability . ] . if on the other hand @xmath591 , then @xmath592 and @xmath593 . * in this case , @xmath594 and there is no update at the buffers of the node @xmath575 .
therefore , @xmath595 * + notice that in this case , since the occupancy of @xmath575 before the update is not full , there exists @xmath596 such that @xmath597 suppose that @xmath598 is used to update @xmath599 with the message @xmath600 .
then , @xmath601\end{array}\in { \mathcal{w}}_{h - j+1}(l , j)}\nonumber\\ & \leftrightarrow ( \sigma_k a_kb_k ) m_{h - j+1}(l ) \in { \mathcal{w}}_{h - j+1}(l , j)\nonumber\end{aligned}\ ] ] + now , note that the vector @xmath551 is not unique and that the set of all vectors that relate the contents of @xmath575 form a vector space over @xmath545 of dimension @xmath602 .
let @xmath566 be the matrix generated by enlisting all the vectors in this space as rows .
then @xmath603 and by cor .
[ linnolemcor ] , @xmath604=\frac{1}{q^{m_{h - j+1}-\eta_{h - j+1}(l , j-1)}}$ ] .
therefore , by choosing a large field size , the probability of the event @xmath605 can be made arbitrarily close to unity .
finally from lemma [ rlcinnolem ] , we see that if @xmath606 , then @xmath600 is innovative w.h.p .
it is straightforward to see that in this setting , if @xmath606 , an innovative packet is conveyed w.h.p . from @xmath574 to @xmath575 and @xmath607 finally , if @xmath608 , we notice that both occupancies remain unaltered w.h.p . * + suppose that by updating the buffers of @xmath575 with @xmath600 ( using a randomly selected @xmath609 ) , we introduce a linear dependency in the newly formed buffer entries . that is , @xmath610 such that @xmath611\end{array}}\in \underbrace{{\mathcal{w}}_{h - j+2}(l , j-1)}_{\subseteq { \mathcal{w}}_{h - j+1}(l , j-1)}\nonumber\\ & \leftrightarrow ( \sigma_k a_kb_k ) m_{h - j+1}(l ) \in { \mathcal{w}}_{h - j+1}(l , j-1)\nonumber\\ & \leftrightarrow(\sigma_k a_kb_k=0)\vee\big(m_{h - j+1}(l ) \in { \mathcal{w}}_{h - j+1}(l , j-1)\big)\nonumber\end{aligned}\ ] ] however , from lemma [ rlcinnolem ] and cor . [ linnolemcor ] , @xmath612<o(\frac{1}{q})$ ] and @xmath613=\frac{1}{q}$ ] . therefore , w.h.p . there is no linear dependency introduced after update and the occupancy is unaltered in this case .
* + just like before , the aim here is to show that there will be no change in occupancy . since in this case
, @xmath574 has no innovative packets , the message it generates will be a linear combination of packets in @xmath614 , @xmath615 .
therefore , we can write @xmath616 , where @xmath617 and @xmath618 . let @xmath598 be used to update the buffer of @xmath575 and let @xmath619 , then , @xmath620\end{array}{\in w_{h - j+2}(l , j)}\nonumber\\ & \leftrightarrow { \sigma_{k , l } \big[a_k + e_ka_l b_l\big]p_{h - j+1,k}(l , j-1 ) \in w_{h - j+2}(l , j)}\nonumber\end{aligned}\ ] ] note that the above is true if only if @xmath621 for @xmath622 , since @xmath623 . therefore
, a linear dependency of stored packets arises if and only if there is a non - trivial solution for @xmath624 , where @xmath625 $ ] , and @xmath626 $ ] .
however , this occurs if and only if @xmath627 .
finally , note that this determinant is zero if and only if @xmath628 .
however , this event occurs with probability @xmath629 , since the vector @xmath609 is chosen uniformly at random from @xmath630 .
therefore , w.h.p . there is no linear dependency induced in the contents of @xmath575 and the occupancy of @xmath575 remains @xmath631 .
* although the dynamics of the system are driven by the spaces @xmath632 , the transitions and their probabilities depend only on @xmath633 , @xmath8 , @xmath462 , and not the spaces as such
. therefore , the system can be equivalently modeled using just these occupancy vectors as states .
* the transition probabilities for the chain given by occupancies @xmath634 , @xmath8 approach that of the emc as the field size is made large .
finally , since the steady - state probability is a continuous function of the probability transition matrix , the steady - state probabilities of the chain for networks without feedback approaches that of the emc , thereby guaranteeing that the throughput achieved by the random coding scheme over a line network without feedback is asymptotically the same as that of a line network with identical parameters and perfect feedback .
a. f. dana , r. gowaikar , and b. hassibi , `` on the capacity region of broadcast over wireless erasure networks , '' _ in proc .
annual allerton conference on communication , control and computing _ , october 2004 .
d. s. lun , n. ratnakar , r. koetter , m. mdard , e. ahmed , and h. lee , `` achieving minimum - cost multicast : a decentralized approach based on network coding , '' _ in proc .
24th annual joint conference of the ieee computer and communications societies , infocom _ , vol . 3 , mar .
2005 .
b. n. vellambi , n. rahnavard , and f. fekri , `` the effect of finite memory on throughput of wireline packet networks , '' _ in proc .
information theory workshop ( itw 2007 ) , lake tahoe , ca _
, september 2007 .
d. s. lun , p. pakzad , c. fragouli , m. mdard , and r. koetter , `` an analysis of finite - memory random linear coding on packet streams , '' _
2nd workshop on network coding , theory , and applications ( netcod 2006 ) , boston , ma , april 3 - 7 , 2006_. | this work investigates the effect of finite buffer sizes on the throughput capacity and packet delay of line networks with packet erasure links that have perfect feedback .
these performance measures are shown to be linked to the stationary distribution of an underlying irreducible markov chain that models the system exactly . using simple strategies , bounds on the throughput capacity
are derived .
the work then presents two iterative schemes to approximate the steady - state distribution of node occupancies by decoupling the chain to smaller queueing blocks .
these approximate solutions are used to understand the effect of buffer sizes on throughput capacity and the distribution of packet delay . using the exact modeling for line networks
, it is shown that the throughput capacity is unaltered in the absence of hop - by - hop feedback provided packet - level network coding is allowed . finally ,
using simulations , it is confirmed that the proposed framework yields accurate estimates of the throughput capacity and delay distribution and captures the vital trends and tradeoffs in these networks .
finite buffer , line network , markov chain , network coding , packet delay , throughput capacity . |
in 2005 , the bes collaboration observed a narrow peak called @xmath0 in the reaction @xmath11 with a statistical significance of @xmath12 , and the mass and width of the @xmath0 obtained from the fit with a breit - wigner function are @xmath13 mev and @xmath14 mev , respectively@xcite .
no partial wave analysis has been made but based on production and decay , it is most likely that the @xmath15 of the @xmath0 is @xmath16 .
the bes collaboration suggests that the @xmath0 is related to the @xmath17 mass threshold enhancement observed in the @xmath18 channel@xcite , which therefore triggering many exotic speculations of the nature of the @xmath0 such as glueball@xcite , @xmath17 baryonium@xcite etc .
it should be noted that there is no strong experimental evidence that the @xmath17 threshold enhancement and the @xmath0 are the same resonance , as pointed out by huang and zhu@xcite ; very probably they have completely different underlying structures . in the presence of the @xmath0 being not related to the @xmath17 mass threshold enhancement , huang and zhu suggested the @xmath0 is the second radial excitation of the @xmath19@xcite , while klempt and zaitsev rather believed the @xmath0 to be the @xmath19 s first radial excitation@xcite .
the @xmath1 was reported by the markiii collaboration in radiative @xmath20 decays into @xmath21@xcite and @xmath22@xcite .
it was also observed by the dm2 collaboration in @xmath20 radiative decays in the @xmath22@xcite decay mode with a mass of @xmath23 mev and a width of @xmath24 mev and in the @xmath21 decay mode@xcite .
vijande et al . suggested that it can be identified as the @xmath25 @xmath10 state@xcite ; page and li proposed that it could have hybrid admixture@xcite . a reanalysis markiii data on @xmath26 performed by bugg et al .
suggested that the decay mode should not be in the @xmath22 pseudoscalar wave but in the @xmath27 @xmath28 scalar wave@xcite .
this was supported by bes collaboration suggesting the @xmath22 resonance at @xmath29 mev should be interpreted as the scalar meson in its @xmath30 decay@xcite .
the conclusions regarding the presence of this pseudoscalar signal in the markiii @xmath31 data become therefore controversial .
more recently , the decay channel @xmath32 was analyzed by the bes collaboration , and a strong enhancement was found in the @xmath21 invariant mass distribution at @xmath33 gev .
the partial wave analysis indicated that the structure was predominantly pseudoscalar with small scalar and tensor contributions .
the mass of the pseudoscalar state also called @xmath1 is @xmath34 mev , the width @xmath35 mev and the product branching fraction is @xmath36@xcite .
the bes collaboration suggested that it could be a mixture of the glueball and the @xmath10 meson . from pdg2006@xcite , the @xmath37 meson nonet ( @xmath38 , @xmath39 , @xmath19 and @xmath40 ) as well as the @xmath25 members [ @xmath41 , @xmath42 and @xmath43 have been well established .
we argue that the @xmath0 and @xmath1 reported by bes collaboration can be assigned as the ordinary @xmath2 @xmath10 states .
the purpose of this work is to check in the simple picture of the @xmath0 and @xmath1 as the @xmath2 @xmath10 states , whether the total widths of these two states can be reasonably reproduced in the @xmath3 meson decay model or not .
the organization of this paper is as follows . in section 2 ,
the brief review of the @xmath3 decay model is given ( for the detailed review see _
e.g. _ refs.@xcite . ) in section 3 , the decay widths of the @xmath0 and @xmath1 are presented .
the mixing angle of the @xmath44 is given in section 4 , and a summary and conclusion are given in section 5 .
the @xmath3 decay model , also known as the quark - pair creation model , was originally introduced by micu@xcite and further developed by le yaouanc et al.@xcite .
the @xmath3 decay model has been widely used to evaluate the strong decays of hadrons@xcite , since it gives a good description of many of the observed decay amplitudes and partial widths of the hadrons .
the main assumption of the @xmath3 decay model is that strong decays take place via the creation of a @xmath3 quark - antiquark pair from the vacuum . the new produced quark - antiquark pair , together with the @xmath10 within the initial meson regroups into two outgoing mesons in all possible quark rearrangement ways , which corresponds to the two decay diagrams as shown in fig.1 for the meson decay process @xmath45 . the transition operator @xmath46 of the decay @xmath47 in the @xmath3 model is given by @xmath48 where @xmath49 is a dimensionless parameter representing the probability of the quark - antiquark pair @xmath50 with @xmath51 creation from the vacuum , @xmath52 and @xmath53 are the momenta of the created quark @xmath54 and antiquark @xmath55 , respectively .
@xmath56 , @xmath57 , and @xmath58 are the flavor , color , and spin wave functions of the @xmath50 , respectively .
the solid harmonic polynomial @xmath59 reflects the momentum - space distribution of the @xmath50 . for the meson wave function
, we adopt the mock meson @xmath60 defined by@xcite @xmath61 where @xmath62 and @xmath63 are the masses of the quark @xmath64 with a momentum of @xmath65 and the antiquark @xmath66 with a momentum of @xmath67 , respectively .
@xmath68 is the radial quantum number of the meson @xmath69 composed of @xmath70 .
@xmath71 , @xmath72 , @xmath73 ( @xmath74 ) is the spin of @xmath64 ( @xmath75 ) , @xmath76 is the relative orbital angular momentum between @xmath64 and @xmath75 .
@xmath77 , @xmath78 .
@xmath79 is a clebsch - gordan coefficient , and @xmath80 is the total energy of the meson @xmath69 .
@xmath81 , @xmath82 , @xmath83 , and @xmath84 are the spin , flavor , color , and space wave functions of the meson @xmath69 , respectively .
the mock meson satisfies the normalization condition @xmath85 the @xmath86-matrix of the process @xmath47 is defined by @xmath87 with @xmath88 where @xmath89 is the helicity amplitude of @xmath47 . in the center of mass frame of meson @xmath69 ,
@xmath89 can be written as @xmath90{l } \scriptstyle m_{l_a},m_{s_a},\\\scriptstyle m_{l_b},m_{s_b},\\ \scriptstyle m_{l_c},m_{s_c},m \end{array}}\renewcommand{\arraystretch}{1}\!\ !
\langle l_am_{l_a}s_am_{s_a}|j_am_{j_a}\rangle\nonumber\\ & & \times\langle l_bm_{l_b}s_bm_{s_b}|j_bm_{j_b}\rangle\langle l_cm_{l_c}s_cm_{s_c}|j_cm_{j_c}\rangle\nonumber\\ & & \times\langle 1m1-m|00\rangle\langle \chi^{14}_{s_bm_{s_b}}\chi^{32}_{s_cm_{s_c}}|\chi^{12}_{s_am_{s_a}}\chi^{34}_{1-m}\rangle\nonumber\\ & & \times[f_1i(\vec{p},m_1,m_2,m_3)+(-1)^{1+s_a+s_b+s_c}f_2i(-\vec{p},m_2,m_1,m_3)],\end{aligned}\ ] ] with @xmath91 and @xmath92 , corresponding to the contributions from figs . 1 ( a ) and 1 ( b ) , respectively , and @xmath93 where @xmath94 , @xmath95 , @xmath96 is the mass of the created quark @xmath54 .
the spin overlap in terms of winger s @xmath97 symbol can be given by @xmath98 in order to compare with experiment conventionally , @xmath99 can be converted into the partial amplitude by a recoupling calculation@xcite @xmath100{l } \scriptstyle m_{j_b},m_{j_c},\\\scriptstyle m_s , m_l \end{array}}\renewcommand{\arraystretch}{1}\!\ ! \langle
lm_lsm_s|j_am_{j_a}\rangle\langle j_bm_{j_b}j_cm_{j_c}|sm_s\rangle\nonumber\\ & & \times\int d\omega\,\mbox{}y^\ast_{lm_l}{\cal{m}}^{m_{j_a}m_{j_b}m_{j_c } } ( \vec{p}).\end{aligned}\ ] ] if we consider the relativistic phase space , the decay width @xmath101 in terms of the partial wave amplitudes is @xmath102 here @xmath103=@xmath104[m^2_a-(m_b - m_c)^2]}}{2m_a}$ ] , @xmath105 , @xmath106 , and @xmath107 are the masses of the meson @xmath69 , @xmath108 , and @xmath109 , respectively . the decay width can be derived analytically if the simple harmonic oscillator ( sho ) approximation for the meson space wave functions is used . in momentum - space ,
the sho wave function is @xmath110 where the radial wave function is given by @xmath111 here @xmath112 is the sho wave function scale parameter , and @xmath113 is an associated laguerre polynomial .
the sho wave functions can not be regarded as realistic , however , they are a _ de facto _ standard for many nonrelativistic quark model calculations .
moreover , the more realistic space wave functions such as those obtained from coulomb plus the linear potential model do not always result in systematic improvements due to the inherent uncertainties of the @xmath3 decay model itself@xcite .
the sho wave function approximation is commonly employed in the @xmath3 decay model in literature . in the present work ,
the sho wave function approximation for the meson space wave functions is taken .
it is well known that in a meson nonet , the physical isoscalar states can mix . with the assumption that the @xmath0 and @xmath1 being the @xmath2 meson nonet , the @xmath44 mixing can be parameterized as @xmath114 where @xmath115 and @xmath116 are the pure @xmath2 nonstrange and strange states , respectively . under the sho wave function approximation ,
the parameters used in the @xmath3 decay model involve the @xmath10 pair production strength parameter @xmath49 , the sho wave function scale parameter @xmath112 , and the masses of the constituent quarks . in the present work ,
we take @xmath118 and @xmath119 gev , the typical values used to evaluate the light meson decays@xcite is higher than that used by other groups such as @xcite by a factor of @xmath120 due to different field conventions , constant factor in @xmath46 , etc .
the calculated results of the widths are , of course , unaffected . ] , and @xmath121 gev , @xmath122 gev@xcite . based on the partial wave amplitudes of the @xmath2 @xmath10 into two other mesons listed in the appendix a and the flavor and charge multiplicity factors shown in table 2 , from ( [ w1 ] ) and ( [ w2 ] ) , the numerical values of the partial decay widths of the @xmath1 and @xmath0 are listed in table 1 .
masses of the final mesons are taken from pdg2006@xcite .
we assume the @xmath123 is the ground scalar mesons as ref.@xcite .
the total widths of the @xmath0 and @xmath1 are shown in fig .
2 as functions of the mixing angle @xmath124 . | with the assumption that the @xmath0 and @xmath1 recently observed by the bes collaboration are the @xmath2 meson states , the strong decays of these two states are investigated in the @xmath3 decay model .
we find that the predicted total widths of the @xmath0 and @xmath1 can be reasonably reproduced with the @xmath4 mixing angle lying on the range from @xmath5 to @xmath6 radians .
further , the mixing angle of the @xmath0 and @xmath1 is phenomenologically determined to be about @xmath7 radians in the presence of the @xmath8 , @xmath9 , @xmath1 and @xmath0 belonging to the @xmath2 meson nonet .
our estimated mixing angle can naturally account for the widths of the @xmath0 and @xmath1 , which shows that the assignment of the @xmath0 and @xmath1 , together the @xmath8 and @xmath9 as the @xmath2 @xmath10 members seems reasonable .
-1 cm mesons , @xmath3 model * pacs numbers:*14.40 .
cs , 12.39.jh |
second - order phase transitions have universal behavior , associated with long wavelength fluctuations , for which critical exponents and other universal quantities can often be successfully calculated using renormalization group techniques . for most such systems
, the short distance physics is hopelessly complicated .
in contrast , the phase transition of a dilute , interacting bose gas provides a fascinating example where physics becomes simpler , and perturbative , at ( relatively ) small distance scales . for this system
, it should be possible to marry techniques for treating long - distance critical fluctuations to a perturbative treatment of short distance physics , and so compute non - universal characteristics of the phase transition .
a simple example of such a non - universal quantity is the phase transition temperature @xmath2 , and the effect of interactions on @xmath2 has been explored by several authors @xcite , with a wide variety of theoretical results .
in particular , the transition temperature has recently been calculated by baym , blaizot , and zinn - justin @xcite in the large @xmath0 approximation . for simplicity ,
they implicitly focus on the case of bose gases with a single spin state , where the low - energy cross - section for atomic collisions can be parametrized by a single scattering length , @xmath3 .
as will be briefly reviewed below , the problem is first reduced to a calculation in a three dimensional o(2 ) scalar field theory at its critical point . replacing that by an o(@xmath0 ) theory with @xmath4
, they find @xmath5 + o\left((a n^{1/3})^2\right ) \right ] \label { eq : tclo}\ ] ] in the dilute limit , where @xmath6 is the number density , is fixed . alternatively , in an arbitrarily wide harmonic trap , @xmath6 should be interpreted as the actual density at the center of the trap at the transition temperature . ] and @xmath7 is the transition temperature of a non - interacting bose gas , @xmath8 their result of @xmath9 is in good agreement with recent in @xmath10 .
there has also been some experimental data on the @xmath11he - vycor system @xcite , which superficially seems to fit well an early theoretical estimate of stoof @xcite , which is @xmath12 .
however , the detailed interpretation of this data is unclear . in that experiment ,
the helium atoms are confined to an interconnected network of channels in the porous vycor glass , and , for the low - density data that appears to fit stoof , the interpaticle spacing is the same order of magnitude as the widths of the channels . ref .
@xcite simply assumes that the system can be modeled by a free bose gas with ( i ) an effective mass for the atoms that is extracted experimentally , but ( ii ) the same scattering length as for bulk helium , which is moreover taken from theoretical modeling . because of these assumptions , the apparent agreement with ref .
@xcite should be treated with caution . ]
numerical simulations @xcite that give @xmath13 .
the result is surprising because it seems to work much better than the large @xmath0 expansion of critical exponents .
for example , for @xmath14 theory , the susceptibility critical exponent @xmath15 is @xcite , whose actual value is @xmath16 for @xmath4 .
the fact that o(2 ) critical exponents should be identified with bose gas exponents is not completely trivial .
a uniform , non - relativistic bose gas is a constrained system : the particle density @xmath6 is fixed .
this constraint causes the critical exponents @xmath17 of the actual system to be related @xcite to the standard exponents @xmath18 of the field theory by ( i ) @xmath19 , and @xmath20 for the others , if @xmath21 , or ( ii ) @xmath22 if @xmath23 .
the actual value of @xmath24 for the o(2 ) model is believed to be @xmath25 @xcite .
if negative , there is no difference between the exponents ; if positive , there is in principle a very tiny difference .
this relation explains , by the way , the difference between mean - field theory exponents for the o(2 ) model ( e.g. @xmath26 ) and the exponents of a non - interacting bose gas ( e.g. @xmath27 ) . ]
@xmath28 this is not a marvelous expansion for @xmath4 , for which the actual value is @xmath29 . in this paper , we calculate the @xmath30 relative correction to ( [ eq : tclo ] ) .
we find @xmath31 + o\left((a n^{1/3})^2\right ) \right ] .
\label { eq : tcnlo}\ ] ] setting @xmath4 , this is only an 26% correction to the leading large @xmath0 result for @xmath10 .
we now have @xmath32 .
though this does not agree as well with the quoted simulation result , the moderatley small size of the correction supports the proposition that the large @xmath0 expansion works surprisingly well for @xmath2 . in the remainder of this introduction ,
we review the long - distance o(2 ) effective theory for bose condensation and then review the arguments of @xcite about how to calculate @xmath33 . in sec . [ sec : lo ] , we review the leading - order calculation in large @xmath0 as done in @xcite . in sec . [ sec : nlo ] , we go on to calculate the next order in @xmath1 .
an appendix explains how to calculate some of the basic 3-dimensional integrals that appear at that order .
the basic assumption throughout will be that the average separation @xmath34 of atoms is large compared to the scattering length @xmath3 .
this can also be expressed as @xmath35 , where @xmath36 is the thermal wavelength @xmath37 it is well known that , at distance scales large compared to the scattering length @xmath3 , an appropriate effective theory for a dilute bose gas is the second - quantized schrdinger equation , together with a chemical potential @xmath38 that couples to particle number density @xmath39 , and a @xmath40 contact interaction that reproduces low - energy scattering .
the corresponding lagrangian is @xmath41 in this context the corresponding mean - field equation of motion is called the gross - pitaevskii equation . as with any effective theory , there are corrections represented by higher - dimensional , irrelevant interactions ( in the sense of the renormalization group ) , and @xmath42 terms of the action @xmath43 determine that times scales as ( length)@xmath44 and the scaling dimension of @xmath45 is ( length)@xmath46 . ] such as @xmath47 and @xmath48 .
however , higher and higher dimension operators are parametrically less and less important if the distance scales of interest are large compared to the characteristic scales ( @xmath3 ) of the atomic interactions .
the @xmath49 term in the lagrangian ( [ eq : l1 ] ) is in fact the lowest - dimension irrelevant interaction , and it is adequate for computing the leading - order effects of interactions in the diluteness expansion .
now treat the system at finite temperature using euclidean time formalism .
the field can then be decomposed into frequency modes with matsubara frequencies @xmath50 . at sufficiently large distance scales ( @xmath51 ) , and small chemical potential ( @xmath52 ) , the @xmath53 terms in ( [ eq : l1 ] ) become small compared to the @xmath54 time derivative term , provided @xmath55 .
the non - zero matsubara frequency modes then decouple from the dynamics , leaving behind an effective theory of only the zero - frequency modes @xmath56 . roughly , @xmath57 \label { eq : l2}\ ] ] with @xmath58 . in detail
, the parameters of ( [ eq : l2 ] ) are renormalized by coupling to the non - zero modes , and there are again corrections in the form of irrelevant ( and even marginal ) interactions .
however , these effects are all suppressed in the dilute limit ) , the short - distance scaling dimension of @xmath56 is ( length)@xmath59 , the @xmath60 interaction is relevant , and a @xmath61 interaction would be marginal .
even though marginal , this last interaction can be ignored at the order of interest in the diluteness expansion because it has a small coefficient .
for example , consider the term that would arise directly from the presence of a correction @xmath62 to the original lagrangian ( [ eq : l1 ] ) .
that would lead to a @xmath63 term in ( [ eq : l2 ] ) which , after rescaling , would become a term proportional to @xmath64 in ( [ eq : o2 ] ) .
since @xmath65 at the transition , the coefficient of this term is high order in the diluteness expansion in @xmath66 .
similarly , an effective @xmath67 term arising from the 4-point interactions @xmath49 and from integrating out physics at the scale @xmath36 ( due , for example , to non - zero matsubara modes ) would give rise to a @xmath68 term in ( [ eq : o2 ] ) with coefficient proportional to @xmath69 . ] and do not affect the computation of @xmath70 at leading order in @xmath71 .
it is then convenient to write @xmath72 so that the effective action @xmath73 becomes a conventionally normalized o(2 ) field theory : @xmath74 , \label{eq : o2}\ ] ] where @xmath75 is understood to be a 2-component real vector @xmath76 and @xmath77 our effective theory depends on two as yet undetermined parameters@xmath78 and @xmath79 , or equivalently @xmath38 and @xmath80 .
one constraint comes from fixing particle number density @xmath6 : [ eq : constraints]@xmath81at the critical temperature , a second requirement is that the system have infinite correlation length , which requires @xmath82 where @xmath83 is the proper self - energy of the @xmath75 field . the two equations ( [ eq : constraints ] ) determine the two unknowns @xmath78 and @xmath79 , and hence @xmath2 . as noted by baym _
_ , the density equation ( [ eq : n0 ] ) can be rewritten as @xmath84 } , \label { eq : n}\ ] ] where the last equality uses ( [ eq : xi ] ) . throughout this paper
, we will use the notational shorthand @xmath85 for momentum integrals .
( technically , @xmath86 is a wave number rather than a momentum , but we will use conventional @xmath87 nomenclature , even though we have not set @xmath88 to 1 . ) the expression ( [ eq : n ] ) for the density is ultraviolet ( uv ) divergent and so receives contributions from short distance scales where the effective theory breaks down .
this could be handled by appropriately regulating the effective theory and then perturbatively correcting the uv contribution . as pointed out by baym _
et al . _ , it is simpler to instead consider the difference @xmath89 , where @xmath90 is the same expression in the absence of interactions ( i.e. , with @xmath91 set to zero ) : @xmath92\over m \kb t } = \int_\p \left[{1\over p^2 + r + \pi(p ) } - { 1\over p^2}\right ] = \int_\p \left[{1\over p^2 + [ \pi(p)-\pi(0 ) ] } - { 1\over p^2}\right ]
. \label { eq : dn}\ ] ] @xmath90 represents the density a _ non_-interacting bose gas has if its transition temperature is @xmath80 .
it is given by inverting ( [ eq : t0 ] ) : @xmath93 this formula can not be derived directly in the effective theory ( [ eq : o2 ] ) , but the difference @xmath94 in ( [ eq : dn ] ) is insensitive to the uv and so can be .
the above constraints are entirely adequate to systematically determine @xmath95 in the large @xmath0 expansion , but there is a convenient way to simplify the bookkeeping a bit .
_ give a simple argument that , to leading order in the density expansion , @xmath96 \over n } \
, , \label { eq : usedn}\ ] ] where the factor of @xmath97 in ( [ eq : dn ] ) arises from the relation @xmath98 . combining ( [ eq : usedn ] ) with ( [ eq : dn ] ) , we can summarize as @xmath99
= - { 2 m \kb t_0\over 3\hbar^2 n } \int_\p \left[{1\over p^2 + [ \pi(p)-\pi(0 ) ] } - { 1\over p^2}\right ] .\ ] ] to leading order in @xmath71 .
it s also useful to rephrase this , again in terms of the fields @xmath75 of the effective theory , as @xmath100 where @xmath101 note that the problem of calculating @xmath102 from the action ( [ eq : o2 ] ) , subject to the constraint ( [ eq : xi ] ) , has only one dimensionful scale in it :
the length scale of this problem , which will be the length scale of the physics that determines @xmath10 , is therefore @xmath103 by dimensional analysis . in the dilute limit @xmath104 , this length scale is large compared to @xmath36 , which justifies use of the o(2 ) effective theory ( [ eq : o2 ] ) .
we now review the leading large-@xmath0 calculation of @xmath105 , and hence of @xmath10 , by baym _
the details of our calculation are slightly different than theirs , and we will introduce techniques needed to proceed to higher order .
we start with the standard large-@xmath0 generalization of the o(2 ) scalar field theory ( [ eq : o2 ] ) to an @xmath14 scalar theory : replace @xmath75 by an @xmath0-component vector and treat @xmath106 as fixed in the @xmath107 limit .
the reader should keep in mind that @xmath79 is therefore order @xmath1 .
standard @xmath0 power counting of feynman diagrams consists of a power of @xmath108 for each 4-point vertex and a power of @xmath0 for each flavor trace .
the set of diagrams that determine @xmath102 at leading order in @xmath1 is depicted in fig .
[ fig : lo ] , where the dashed line denotes bubble chains , as shown in fig .
[ fig : chain ] .
( for comparison , the diagram for @xmath109 is shown in fig .
[ fig:0 ] . )
the cross denotes an insertion of the operator @xmath110 , whose expectation we are computing .
there is a simple way to summarize the effect on diagrammatic perturbation theory of the @xmath111 term in the action ( [ eq : o2 ] ) and the constraint ( [ eq : xi ] ) that @xmath112 . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ rule 1 : use massless ( gapless ) scalar propagators @xmath113 when evaluating diagrams , ignoring the @xmath114 term in the action .
but whenever there is a one - particle irreducible sub - diagram @xmath115 that represents a contribution to the @xmath75 proper self - energy @xmath83 , then .
it is potentially ambiguous for calculating the free energy for example , a diagram like fig .
[ fig : lo ] but without the cross on it . in that case , it is ambiguous which sub - diagrams would be considered self - energy insertions . a systematic way to treat
the perturbation theory in all cases is to treat the @xmath111 term in the action as a perturbation , include it in feynman diagrams as a 2-point vertex , and then set @xmath112 order by order in perturbation theory .
] replace @xmath116 by @xmath117 .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we note for later reference that , for the purpose of this rule , a diagram that is cut in two pieces only by cutting a single internal _ dashed _ line is still one - particle irreducible , because cutting the bubble chain represented by a dashed line corresponds to cutting two @xmath75 lines ( fig .
[ fig : chain ] ) . the bubble chain sum shown in fig .
[ fig : chain ] is given by @xmath118 where [ eq : fp ] @xmath119 and @xmath120 represents the basic massless bubble integral ( not summed over flavors ) @xmath121 in @xmath122 dimensions , @xmath123 putting everything together , the diagram of fig .
[ fig : lo ] gives @xmath124 + o(n^{-1 } ) .
\label{eq : loint}\ ] ] as pointed out by baym _
_ in @xcite , the above integral is not absolutely convergent in three dimensions , and one must be careful to consistently regulate the theory before proceeding .
integrals that are not absolutely convergent are at best ambiguous they depend on the order one chooses to do the integrations .
for example , if one evaluates ( [ eq : loint ] ) directly in three dimensions , doing the angular integrations first , then the @xmath125 integration , and then the @xmath86 integration , the result is zero .
this is not in fact the correct answer .
we will discuss this issue in some detail in order to justify the correctness of our procedure for later evaluating higher - order diagrams .
baym _ et al .
_ s preferred method for the leading - order calculation is to use dimensional regularization and evaluate everything in @xmath126 dimensions .
this is difficult at next order in @xmath1 : the loop integrals we shall encounter are sufficiently complicated that evaluation in @xmath126 dimensions seems hard .
our strategy will be to instead always reduce diagrams to well - defined three - dimensional integrals , which are simpler to evaluate .
we imagine starting with some consistent regularization scheme , like dimensional regularization , and will now discuss how to manipulate the integrals so that they will be absolutely convergent if we set @xmath122 .
we assume in what follows that the uv regulator respects parity and is invariant under shifts @xmath127 of loop momenta @xmath128 .
let s look at the divergences that cause absolute convergence of the integral ( [ eq : loint ] ) to fail in three dimensions .
a simple one to correct is the behavior for @xmath129 fixed and @xmath130 .
the large @xmath86 piece of the @xmath128 integration then behaves as @xmath131 , which is logarithmically uv divergent ( from the point of view of absolute convergence ) .
this can be remedied by rewriting the _ regulated _ version of ( [ eq : loint ] ) by using the freedom to change the integration variable @xmath128 to @xmath132 : @xmath133 .
\label{eq : loint2}\ ] ] now , if we throw away the uv regulator , the @xmath129 fixed , large @xmath86 divergence is gone .
this sort of divergence is trivial , easy to remedy , and wo nt have much practical impact on our calculations ( given the order in which we will eventually do integrations ) .
we ll simply acknowledge the issue in later calculations , without emphasizing it , by writing ( [ eq : loint2 ] ) as @xmath134_\pm , \label { eq : loint2b}\ ] ] where the subscript @xmath135 means that one should average the expression with @xmath136 ( or equivalently with @xmath137 ) . unfortunately , even ( [ eq : loint2 ] ) is not absolutely convergent .
there is still a logarithmic uv divergence associated with @xmath129 and @xmath128 _ simultaneously _ becoming large ( @xmath138 ) , as can be seen by simple power counting and the fact that @xmath139 approaches a non - zero constant for large @xmath86 .
return to considering ( [ eq : loint2 ] ) with a uv regulator still in place .
we can eliminate the uv divergence by rewriting @xmath140_\pm - \int_{\l\p } { \f_\infty\over l^4 } \left[{1\over|\l+\p|^2 } - { 1\over p^2}\right]_\pm .\ ] ] the second integral vanishes , as can be seen by changing integration variable @xmath141 in its first term .
so @xmath140_\pm .
\label { eq : lob}\ ] ] this is now uv convergent because @xmath142 as @xmath130 .
however , we have traded the logarithmic uv divergence for a logarithmic infrared ( ir ) divergence , associated with @xmath143 .
we now need some sort of infrared regulator .
one physically motivated possibility for consistently regulating the infrared would be to consider the system infinitesimally above the critical temperature , so that all the massless scalar propagators @xmath113 should be replaced by massive ones @xmath144 , where the mass @xmath145 represents the inverse correlation length @xmath146 .
this defines an absolutely convergent integral in 3 dimensions , and the limit @xmath147 would be taken only after the integrations .
massless propagators @xmath113 will be much easier to deal with , however , in higher - order calculations . as a practical matter for computing diagrams , we prefer to introduce as few massive propagators as possible
. it would be convenient , for example , to ir regulate ( [ eq : lob ] ) by introducing @xmath145 only in the @xmath148 propagators : @xmath149_\pm , \label { eq : loreg}\ ] ] where @xmath139 is still defined in terms of the massless bubble integral , as in ( [ eq : fp ] ) .
one might worry that an _ ad hoc _ procedure of putting masses only on some propagators could be inconsistent , so let us argue more carefully .
return to the uv regulated version of ( [ eq : loint2 ] ) and note that the integral is not sensitive to the the region of integration where @xmath125 is infinitesimal , because this particular integral is ir convergent . there
s then no reason we ca nt modify the infrared behavior of the integrand for infinitesimal @xmath125 , without affecting the integral .
so , for instance , @xmath150_\pm .\ ] ] but now , again rewriting @xmath151 , the same steps as before reproduce ( [ eq : loreg ] ) . now that we have an absolutely convergent integral ( [ eq : loreg ] ) , we can do the integration in three dimensions and in any order we choose .
it s convenient to do the @xmath129 integral first : @xmath152_\pm = { 1\over 8\pi m ( p^2+m^2 ) } - { 1\over 8\pi m p^2 } = - { m \over 8\pi p^2 ( p^2+m^2 ) } \,.\ ] ] the `` @xmath135 '' prescription makes no difference to this particular integral , because the @xmath129 integration by itself is completely convergent without it .
note that naively setting @xmath145 to zero at this stage would give the incorrect , zero result mentioned earlier .
instead , we have @xmath153 the overall factor of @xmath145 in the numerator is canceled by a linear ir divergence in the @xmath128 integration , which is cut off by @xmath145 . for small @xmath145 , the integral ( [ eq : plup ] )
is dominated ) is dominated by arbitrarily small @xmath154 .
they may worry because at sufficiently small momentum our perturbative propagators are no longer good approximations to the full propagators .
the _ full _
scalar propagators , for example , actually scale like @xmath155 rather than @xmath148 at small @xmath125 ( @xmath156 ) , where the critical exponent @xmath157 is @xmath30 .
the difference becomes significant when @xmath158 $ ] .
one might worry that the sensitivity of ( [ eq : plup ] ) to @xmath159 is a sign that naive large @xmath0 perturbation theory must break down .
it is important to realize , in the present case , that this infrared sensitivity is simply an artifact of our mathematical manipulations on the infrared - safe expression ( [ eq : loreg ] ) . regardless of whether one used some sort of infrared - improved propagator in ( [ eq : loreg ] ) , that expression is not sensitive to far - infrared momenta .
it is sensitive to momenta @xmath160 , for which there is nothing wrong with a large-@xmath0 expansion based on perturbative propagators .
] by @xmath161 .
so , in the limit of @xmath147 , we can simplify the calculation slightly by replacing @xmath162 by @xmath163 .
so @xmath164 when combined with the formula ( [ eq : tceq ] ) for @xmath95 , this reproduce baym _
s leading large @xmath0 result ( [ eq : tclo ] ) , in which @xmath0 has been set to 2 .
the diagrams which contribute to @xmath102 at next order in @xmath1 are shown in figs . [
fig : nlo ] and [ fig : sigma ] .
the diagrammatic expansion comes from the standard introduction of an auxiliary field @xmath165 , represented by the dashed lines . ,
see , for example , section 2.1 of chapter 8 of @xcite . some people might prefer to replace @xmath165 by @xmath166 in the action ( [ eq : lsig1 ] ) , so that the euclidean path integral for @xmath165 is convergent , but it matters not at all for the purpose of large @xmath0 perturbation theory . ] the @xmath14 action of ( [ eq : o2 ] ) is rewritten as@xmath167 .
\label { eq : lsig1}\ ] ] the @xmath165 propagator is then turned into the bubble chain of fig .
[ fig : chain ] by resumming the basic massless bubble of fig .
[ fig : bubble ] into the @xmath165 propagator .
technically , this is accomplished by trivially rewriting ( [ eq : lsig1 ] ) as [ eq : s ] @xmath168 , \\
s_{\rm subtractions } & = & \int_\p \left [ \half \ , r \phi_{-\p } \phi_\p + \half \ , \sigma_{-\p } \ , n\tilde\sigma_0(p ) \ , \sigma_\p \right ] , \end{aligned}\ ] ] with @xmath139 and @xmath169 given by ( [ eq : fp ] ) .
the terms designated @xmath170 may be ignored if one follows the previous rule 1 as well as _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ rule 2 : do not include any diagrams that have the one - loop bubble , fig .
[ fig : bubble ] , as a sub - diagram .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ note that rule 1 eliminates any tadpole sub - diagrams , such as fig .
[ fig : tadpole ] .
formal large @xmath0 counting of diagrams is simply to count a factor of @xmath171 for each @xmath165 propagator and a factor of @xmath0 for each @xmath75 loop .
the important momentum scale of the problem will be the scale @xmath172 , where the @xmath165 propagator ( [ eq : chain ] ) makes the transition from its small @xmath86 behavior ( @xmath173 ) to its large @xmath86 behavior ( @xmath174 ) .
some authors like to completely integrate @xmath75 out of ( [ eq : lsig1 ] ) , but we prefer to retain it , as there is then a more transparent relationship between feynman diagrams and the corresponding feynman integrals . in evaluating the diagrams of fig .
[ fig : nlo ] , we shall borrow techniques from ref .
@xcite , where somewhat related diagrams were evaluated in gauge theories with large numbers of scalars . our strategy will be to do the @xmath75 loop integrals first , and then tackle the remaining integrals associated with @xmath165 propagators .
let s start with fig .
[ fig : nlo]a .
the corresponding integral is @xmath175_\pm \left [ { 1\over |\l+\q|^2 } - { 1\over q^2 } \right]_\pm .
\label { eq : a}\ ] ] as written , this integral is absolutely convergent and can be evaluated , without regularization , directly in three dimensions . to do the @xmath129 integration ,
however , we find it convenient to temporarily introduce an ir regulator mass @xmath145 .
we may then separately integrate each of the terms of the integrand , which are not individually ir convergent .
we can also use @xmath145 as a trick for reducing powers of @xmath176 . specifically , we rewrite the @xmath129 integral as the @xmath147 limit of @xmath177_\pm \left [ { 1\over |\l+\q|^2 } - { 1\over q^2 } \right]_\pm \nonumber\\ & & \qquad = { 1\over2 } \ , { d^2\over d(m^2)^2 } \ , \int_\l { 1\over ( l^2+m^2 ) } \left [ { 1\over |\l+\p|^2 } - { 1\over p^2 } \right]_\pm \left [ { 1\over |\l+\q|^2 } - { 1\over q^2 } \right]_\pm \nonumber\\ & & \qquad % = \left [ \half\ , { d^2\over d(m^2)^2 } \ , % \left \ { i_1(\p,\q;m ) - p^{-2 } j_1(\q;m ) - q^{-2 } j_1(\p;m ) \right \ } % + \frac{1}{32\ , \pi\ , m^3\ , p^2\ , q^2 } \right]_\pm , = { 1\over2 } \ , { d^2\over d(m^2)^2 } \ ,
\left [ i_1(\p,\q;m ) - p^{-2 } j_1(\q;m ) - q^{-2 } j_1(\p;m ) + p^{-2 } q^{-2 } \left(- { m \over 4\pi } \right ) \right]_\pm , \nonumber\\ & & \label { eq : linta}\end{aligned}\ ] ] where @xmath178 @xmath179 and is @xmath180 plus an @xmath145-independent uv divergence , and @xmath181 derivatives of the latter vanish .
it s of course not necessary to introduce @xmath182 and this spurious uv divergence ; one could simply evaluate @xmath183 directly .
but we find it convenient to consolidate the treatment of such integrals with that of the other terms in ( [ eq : linta ] ) . ]
@xmath184 the integral @xmath185 is straightforward to evaluate .
a particularly simple way to evaluate @xmath186 is to make a conformal transformation which reduces it to the form of @xmath185 .
the results of both integrals , and the conformal transformation between them , are discussed in appendix [ app : integrals ] .
all we need here are the small @xmath145 expansions of those results , which turn out to be @xmath187 @xmath188 putting everything together , @xmath177_\pm
\left [ { 1\over |\l+\q|^2 } - { 1\over q^2 } \right]_\pm \nonumber\\ & & \qquad = \left[{\hat\p\cdot\hat\q \over 8\pi m p^3 q^3 } + { 3(\hat\p\cdot\hat\q)^2 - 1 \over 16 p^3 q^3 |\p-\q|}\right]_\pm
+ o(m ) \nonumber\\ & & \qquad = \left[{3(\hat\p\cdot\hat\q)^2 - 1 \over 16 p^3 q^3 |\p-\q|}\right]_\pm
+ o(m ) .\end{aligned}\ ] ] we can now set @xmath189 .
all that will matter in the integral ( [ eq : a ] ) is the average @xmath190 over the angle between @xmath128 and @xmath191 , which is @xmath192_\pm \left [ { 1\over |\l+\q|^2 } - { 1\over q^2 } \right]_\pm \right\rangle_\theta = { 1\over 40 p_>^6 p _ < } , \ ] ] where @xmath193 we are left with @xmath194 the remaining integrals are easy to do , with the result @xmath195 fig .
[ fig : nlo]b corresponds to @xmath196 @xmath197 - { 1\over p^4 } \left[{1\over|\p+\q|^2 } - { 1\over q^2}\right ] \right\}_\pm .\ ] ] this contribution to @xmath102 is again absolutely convergent if the subscript @xmath135 is taken to mean averaging over @xmath198 and also over @xmath199 .
it is convenient to now rewrite the @xmath129 integral as the @xmath200 limit of @xmath201 - { 1\over ( p^2+m_2 ^ 2)^2 } \left[{1\over|\p+\q|^2 } - { 1\over q^2}\right ] \right\}_\pm \nonumber\\ & & \qquad = { d\over d(m_1 ^ 2 ) } \ , { d\over d(m_2 ^ 2 ) } \biggl\ { i_2(\p+\q,\q , m_1,m_2 ) - { 1\over q^2 } \ , j_2(\p;m_1,m_2 ) \nonumber\\ & & \hspace{15em } - { 1\over(p^2+m_2 ^ 2 ) } \left[{1\over|\p+\q|^2 } - { 1\over q^2}\right ] \left(- { m_1\over 4\pi}\right ) \biggr\}_\pm , \nonumber\\ & & \end{aligned}\ ] ] where @xmath202 @xmath203 the results for @xmath204 and @xmath205 , and their small @xmath206 expansions , are given in appendix [ app : integrals ] .
the final result for the @xmath129 integration , after taking the @xmath207 limit , is @xmath208_\pm , \ ] ] with angular average @xmath209 where @xmath210 is the step function ( 1 for @xmath211 ; 0 for @xmath212 ) .
the remaining integrals are easy to do , giving @xmath213 fig .
[ fig : nlo]c can be evaluated as the others , but the final integrals are a bit more complex .
the diagram gives @xmath214 @xmath215_\pm .\ ] ] the @xmath129 integration can be performed using methods similar to before : @xmath216_\pm , \ ] ] @xmath217 @xmath218 can be reduced to the basic integrals @xmath186 and @xmath185 encountered previously by rewriting the numerator 1 in ( [ eq : h ] ) as @xmath219\ ] ] and then expanding the integrand into the corresponding four terms : @xmath220 .\ ] ] using the expansion ( [ eq : i1expand ] ) of @xmath186 , one obtains @xmath221_{\pm } .\ ] ] this simplifies , after applying the @xmath222_\pm$ ] prescription , to @xmath223 .\ ] ] angular averaging yields @xmath224 , \ ] ] where @xmath225 the remaining integrals over @xmath128 and @xmath191 are no longer so trivial .
notice first that for an arbitrary function @xmath226 one can rewrite @xmath227 so that @xmath228 .\ ] ] this is easiest to evaluate numerically , giving @xmath229 where @xmath230 .
we also have an analytic result : term .
we change variables from @xmath231 to @xmath232 .
this turns @xmath233 into @xmath234 .
@xmath235 can be written as a sum of terms of the form @xmath236 , and the change of integration variable makes the rest of the integrand a rational function of @xmath237 .
we then split this rational function apart by partial fractions and do each integral , yielding di- and tri - logarithms of various arguments .
finally , we use a zoo of polylogarithm identities @xcite to simplify the answer . ] @xmath238 - { 2\over3 } \ , l(3,\chi_8^ { } ) , \ ] ] where [ eq : dirichletl ] @xmath239is a particular case of dirichlet s @xmath240-function , and @xmath241 fig .
[ fig : nlo]d corresponds to @xmath242 @xmath243_{\pm\l } \nonumber\\ & = & - i_1(\p,\q;0 ) \lim_{m\to0 } { d\over d(m^2 ) } \left [ i_1(\p,\q , m ) - p^{-2 } q^{-2 } \left(- { m \over 4\pi}\right ) \right ] \nonumber\\ & = & { \hat\p\cdot\hat\q \over 64 p^3 q^3 |\p-\q|^2 } \,.\end{aligned}\ ] ] here the subscript @xmath244 means we implicitly average over @xmath137 for absolute convergence . doing the remaining integrals by brute force , we find @xmath245 , \ ] ] where @xmath246 is the riemann zeta function .
it s interesting to note that @xmath247 can also be written as @xmath248 . the final class of diagrams , fig .
[ fig : nlo]e , correspond to the leading order diagram , fig .
[ fig : lo ] , with the replacement @xmath249(-n^{-1}\ , \f_\p ) \equiv -n^{-1}\ , { \cal f}_\p , \label { eq : fpsub}\ ] ] where @xmath250 represents the contribution to the @xmath165 self - energy at next - to - leading order , shown in the diagrams of fig .
[ fig : sigma ]
. making this substitution in the leading - order calculation ( [ eq : lo ] ) gives @xmath251 the nice feature of this relation is that we only need to calculate @xmath250 in the small and large @xmath86 limits .
the first self - energy diagram in fig .
[ fig : sigma ] contributes @xmath252 for small @xmath128 , the integration is dominated by @xmath253 and @xmath254 .
so we can ignore @xmath255 , @xmath256 , and @xmath128 compared to @xmath191 and write , the result is @xmath257 $ ] , where @xmath258 , and @xmath259 is the dilogarithm function .
one may double check that the @xmath260 limit agrees with ( [ eq : sig10 ] ) . ]
@xmath261 ^ 2 = - { nu \over 48\pi^2 p^2 } .
\label{eq : sig10}\ ] ] this diagram has a quadratic ir divergence for @xmath262 . in contrast , the other two diagrams only diverge as linear@xmath263log when @xmath262 , and so behave as @xmath264 for small @xmath86 . in summary , @xmath265 for small @xmath128 .
one may check by power counting diagrams and sub - diagrams that @xmath266 . then @xmath267 ande . ]
@xmath268 we conclude by mentioning one technical subtlety , glossed over above , concerning absolute convergence . the integration corresponding to the substitution ( [ eq : fpsub ] ) in the leading order analysis is @xmath269_\pm .\ ] ] in contrast to the analogous leading - order expression ( [ eq : loint2b ] ) , this integral is not absolutely convergent in the infrared ( @xmath270 ) , though it is convergent in the uv .
one might therefore worry about the _
ad hoc _ introduction of an ir regulator @xmath145 in the calculation of this graph .
however , this worry is easily bypassed by rewriting @xmath271_\pm , \label{eq : euv}\ ] ] which should be understood as regulated in the uv .
the uv - regulated integral of the @xmath272 factor vanishes .
( [ eq : euv ] ) is now convergent in the ir , but logarithmically divergent in the uv , just as the original leading - order integral ( [ eq : loint2b ] ) was .
one can now follow through the same argument as in the leading - order case to introduce an ir regulator and then remove the uv divergence , where @xmath139 in the leading - order analysis in now replaced by @xmath273 .
the result is still ( [ eq : e ] ) .
summing all the diagrams then yields the total nlo contribution : @xmath274 , \ ] ] with @xmath275 given by ( [ eq : dirichletl ] ) . combining with the leading - order result ( [ eq : lo ] ) , @xmath276 which is the relative nlo correction we presented for @xmath2 in ( [ eq : tcnlo ] ) .
we are indebted to eric braaten for suggesting this project .
we are also grateful to tim newman and genya kolemeisky for useful discussions .
this work was supported , in part , by the u.s .
department of energy under grant no .
de - fg02 - 97er41027 .
let s begin with the integral @xmath277 .
this is quite easy to do by standard methods ( for example , by introducing a feynman parameter ) , and gives @xmath278 the integral @xmath279 of ( [ eq : j1def ] ) is simply the special case @xmath280 .
the integral @xmath281 of ( [ eq : i2def ] ) can be related to @xmath282 by generalizing a trick presented in ref .
the idea is to change integration variables from @xmath129 to its conformal inversion @xmath283 .
the integration measure changes as @xmath284 .
propagators can be written in terms of the new variable @xmath285 as @xmath286 } \,,\end{aligned}\ ] ] where @xmath287 making this change of variables , @xmath288^{-1 } \nonumber\\ & & \hspace{10em } \times \left [ \left| \tilde\l + { \q\over q^2+m_2 ^ 2}\right|^2 + \left(m_2\over q^2 + m_2 ^ 2\right)^2 \right]^{-1 } \nonumber\\ & = & { 1\over ( p^2+m_1 ^ 2 ) ( q^2+m_2 ^ 2 ) } \ , j_2\!\left [ { \p \over p^2+m_1 ^ 2 } - { \q\over q^2+m_2 ^ 2 } ; { m_1\over p^2+m_1 ^ 2 } , { m_2\over q^2+m_2 ^ 2 } \right ] .\end{aligned}\ ] ] for the application of this paper , the relevant terms in the small @xmath206 expansion of @xmath204 are @xmath289 \over 8 p^2 q^2 |\p-\q|^5 } + o(m^5 ) .\end{aligned}\ ] ] j.c . le guillon and j. zinn - justin , phys . rev .
b * 21 * , 3976 ( 1980 ) ; phys . ref .
lett . * 39 * , 95 ( 1977 ) ; g.a .
baker , b.g .
nickel , d.i .
meiron , phys .
b * 17 * , 1365 ( 1978 ) .
baker , b.g .
nickel , m.s .
green , d.i .
meiron , phys .
lett . * 36 * , 1351 ( 1976 ) .
lee and c.n .
yang , phys . rev . * 105 * , 1119 ( 1957 ) ; t.d .
lee , k. huang , and c.n .
yang , phys .
rev . * 106 * , 1135 ( 1957 ) ; t.t .
wu , phys .
rev . * 115 * , 1390 ( 1959 ) ; n.m .
hugenholtz and d. pines , phys .
rev . * 116 * , 489 ( 1959 ) ; k. sawada , phys .
rev . * 116 * , 1344 ( 1959 ) .
l. lewin , _ polylogarithms and associated functions _
( elsevier : new york , 1981 ) [ or the older version , l. lewin , _ dilogarithms and associated functions _ ( macdonald : london , 1958 ) ] ; l. lewin , ed .
, _ structual properties of polylogarithms _ , mathematical surveys and monographs , v. 37 ( american mathematical society : providence , 1991 ) ; d. zagier in _ arithmetic algebraic geometry _ , ed .
f. van der greet _
( birkhser boston : boston , 1991 ) , pp .
391 - 430 . | baym , blaizot , and zinn - justin have recently used the large @xmath0 approximation to calculate the effect of interactions on the transition temperature of dilute bose gases .
we extend their calculation to next - to - leading - order in @xmath1 and find a relatively small correction of -26% to the leading - order result .
this suggests that the large @xmath0 approximation works surprisingly well in this application .
*l * *p * -20pt |
@xmath103=\kappa \left ( n_{1}\left(t_{2}\right)+1 \right)\left ( 2 a \rho a^{\dagger } -a^{\dagger } a \rho -\rho a^{\dagger } a \right ) + \kappa n_{1}\left(t_{2}\right)\left ( 2 a^{\dagger } \rho a -a a^{\dagger } \rho- \rho a a^{\dagger } \right)\\ \shoveleft{+\gamma_{1}\left ( n_{1}\left(t_{2}\right ) + 1 \right)\sum_{j}\left ( 2 \ket{l_{j}}\bra{u_{j } } \rho \ket{u_{j}}\bra{l_{j}}\right . }
- \ket{u_{j}}\bra{u_{j}}\rho - \rho \ket{u_{j}}\bra{u_{j}}\right)\\ \shoveleft{+\gamma_{1}\left ( n_{1}\left(t_{2}\right ) \right)\sum_{j}\left ( 2 \ket{u_{j}}\bra{l_{j } } \rho \ket{l_{j}}\bra{u_{j}}\right . }
- \ket{l_{j}}\bra{l_{j}}\rho - \rho \ket{l_{j}}\bra{l_{j}}\right)\\ \shoveleft{+\gamma_{2}\left ( n_{2}\left(t_{1}\right ) + 1 \right)\sum_{j}\left ( 2 \ket{i_{2,j}}\bra{i_{3,j } } \rho \ket{i_{3,j}}\bra{i_{2,j}}\right . }
- \ket{i_{3,j}}\bra{i_{3,j}}\rho - \rho \ket{i_{3,j}}\bra{i_{3,j}}\right)\\ \shoveleft{+\gamma_{2}\left ( n_{2}\left(t_{1}\right ) \right)\sum_{j}\left ( 2 \ket{i_{3,j}}\bra{i_{2,j } } \rho \ket{i_{2,j}}\bra{i_{3,j}}\right . }
- \ket{i_{2,j}}\bra{i_{2,j}}\rho - \rho \ket{i_{2,j}}\bra{i_{2,j}}\right)\\ \shoveleft{+\gamma_{3}\left ( n_{3}\left(t_{2}\right ) + 1 \right)\sum_{j}\left ( 2 \ket{u_{j}}\bra{i_{4,j } } \rho \ket{i_{4,j}}\bra{u_{j}}\right . }
- \ket{i_{4,j}}\bra{i_{4,j}}\rho - \rho \ket{i_{4,j}}\bra{i_{4,j}}\right)\\ \shoveleft{+\gamma_{3}\left ( n_{3}\left(t_{2}\right)\left(t_{2}\right ) \right)\sum_{j}\left ( 2 \ket{i_{4,j}}\bra{u_{j } } \rho \ket{u_{j}}\bra{i_{4,j}}\right . }
\left . - \ket{u_{j}}\bra{u_{j}}\rho - \rho \ket{u_{j}}\bra{u_{j}}\right)\\ \shoveleft{+\gamma_{4}\left ( n_{4}\left(t_{2}\right ) + 1 \right)\sum_{j}\left ( 2 \ket{i_{1,j}}\bra{l_{j } } \rho \ket{l_{j}}\bra{i_{1,j}}\right . }
- \ket{l_{j}}\bra{l_{j}}\rho - \rho \ket{l_{j}}\bra{l_{j}}\right)\\ \shoveleft{+\gamma_{4}\left ( n_{4}\left(t_{2}\right ) \right)\sum_{j}\left ( 2 \ket{l_{j}}\bra{i_{1,j } } \rho \ket{i_{1,j}}\bra{l_{j}}\right . }
. - \ket{i_{1,j}}\bra{i_{1,j}}\rho - \rho \ket{i_{1,j}}\bra{i_{1,j}}\right)\ .\end{gathered}\ ] ] for the quantity @xmath62 describing cooperative behavior between the structures .
the stable solution is obtained from the part of eq . which is enclosed by the rectangular brackets .
to find the number of structures necessary for reaching the onset of cooperative behavior @xmath57 for a given temperature @xmath51 we solve @xmath111 to find | a laser can be understood as thermodynamic engine converting heat to a coherent single mode field close to carnot efficiency . from this perspective
spectral shaping of the excitation light generates a higher effective temperature on the pump than on the gain transition . here , using a toy model of a quantum well structure with two suitably designed tunnel - coupled wells kept at different temperature , we study a laser operated on an actual spatial temperature gradient between pump and gain region .
we predict gain and narrow band laser emission for a sufficient temperature gradient and resonator quality .
lasing appears concurrent with amplified heat flow and points to a new form of stimulated solid state cooling .
such a mechanism could raise the operating temperature limit of quantum cascade lasers by substituting phonon emission driven injection , which generates intrinsic heat , by an extended model with phonon absorption steps .
= 1 lasing is facilitated by implementing gain within an optical resonator .
for stimulated emission based amplification this requires an inverted population distribution between the amplifier quantum states @xcite . in theoretical descriptions pumping to generate inversion can be well described by coupling reservoirs of different temperature to the pump , the injection and the lasing transition @xcite .
practically this requires precise spectral design of the pump radiation to overlap mainly with the pump transition with strong suppression on the lasing and injection transitions to yield inversion .
interestingly , such inversion can be consistently mimicked by simply assuming an effective negative temperature on the lasing transition @xcite . at optical frequencies
almost no thermal photons are present at room temperature and simple filtering of the pump light is sufficient . however , simple spectral shaping in the far infrared or even thz regime is practically hampered by ubiquitous thermal quanta , which create strong restrictions on the laser operating temperature @xcite .
in this paper we study possibilities to directly use spatial temperature gradients instead of spectral pump design to generate inversion and facilitate gain and lasing .
although we mostly concentrate on the basic principles and mechanisms of such a setup , we also try to connect to a possible real world implementation based on designed quantum well structures . here
the optical properties and level structures in far infrared or even thz regime can be precisely engineered .
while such a temperature gradient laser at first seems more of conceptual than of practical importance , the underlying principle can be easily generalized to a system with different pump steps . in particular thermal or phonon
absorption based injection steps to the gain transition lead to interesting modification of the thermal operation conditions of such a laser .
lasing can be tied to thermal absorption and heat flow to avoid unwanted heating and control heat flow @xcite .
the model we study is a simple structure comprised of two quantum wells coupled to a cavity .
a sketch is shown in fig .
[ fig : wells ] .
the hamiltonian describing a system of @xmath0 two - well structures coupled to a single cavity mode is given by @xmath1 with @xmath2 \label{eq : hamiltonian0}\end{gathered}\ ] ] and the interaction hamiltonian @xmath3 the energy eigenstates are denoted by @xmath4 , @xmath5 and @xmath6 with the eigenenergies @xmath7 , @xmath8 and @xmath9 and @xmath10 is the creation operator for a photon in the cavity mode with frequency @xmath11 . the coupling between the @xmath12 transition and the mode is described by the coupling rate @xmath13 .
tunneling between the states @xmath14 and @xmath15 and the states @xmath16 and @xmath17 is controlled by the tunneling matrix elements @xmath18 and @xmath19 . for the forthcoming analysis
we define all frequencies with respect to @xmath20 and switch to a frame rotating with @xmath11 introducing the detuning @xmath21 . and @xmath19.,scaledwidth=40.0% ]
the dynamics of the reduced density matrix is described by the master equation @xmath22+\mathcal{l}\left [ \rho \right ] \label{eq : master}\end{aligned}\ ] ] where the liouvillian @xmath23 describes dissipative processes , including the coupling to the environment at finite temperature , see supplementary material .
the decay rate of the cavity is denoted by @xmath24 , while the spontaneous emission rates of the remaining transitions are denoted by @xmath25 .
the thermal photon number at the transition frequencies @xmath26 , @xmath27 , @xmath28 and @xmath29 at temperature @xmath30 is denoted by @xmath31 ( @xmath32 ) .
we assume that we have a temperature gradient across a sample containing @xmath0 two - well structures and that @xmath33 in each structure .
as @xmath34 we will find population in the state @xmath35 . via the tunnel - coupling @xmath18 the state @xmath36 gets populated as well . due to the lower temperature @xmath37 of the second well the population
is preferably transferred to the state @xmath38 . with a certain probability a photon
is created in the cavity and leaves it .
the cycle ends with the tunneling from state @xmath39 to @xmath36 . using eq .
we calculate operator expectation value equations for the populations in each state and the quantities they couple to @xcite .
the equations are truncated to include only second order correlation functions .
the population of the state @xmath5 is described by @xmath40 where we omitted the temperature dependence of the @xmath41 and used the assumption of total phase invariance ( @xmath42 ) to simplify the expansion of third order expectation values @xcite .
this leads to the next equation @xmath43\left
< \ket{u_{j}}\bra{u_{j}}\right>-\left < a^{\dagger}a\right > \left < \ket{l_{j}}\bra{l_{j}}\right > \right)\\ -\left(\kappa+\gamma_{1}\left ( 2n_{1}+1 \right)+\gamma_{3}n_{3}+\gamma_{4}\left ( n_{4}+1 \right ) \right)\left < a \ket{u_{j}}\bra{l_{j } } \right>\\ -\mathrm{i}g\left
( n-1 \right)\left < \ket{l_{j}}\bra{u_{j}}\ket{u_{k}}\bra{l_{k}}\right > \label{eq : au_jl_j}\end{gathered}\ ] ] which describes the rate at which excitations are transferred between the cavity and the structure .
the number of photons follows @xmath44 . in the last line of eq .
we see that cooperative effects between the structures emerge as a result of the common interaction with the cavity .
we find @xmath45\\ -\left ( 2 \gamma_{1}\left ( 2 n_{1}+1 \right)+2 \gamma_{3}n_{3}+2 \gamma_{4}\left ( n_{4}+1 \right ) \right)\\ \times\left < \ket{l_{j}}\bra{u_{j}}\ket{u_{k}}\bra{l_{k}}\right>\ .
\label{eq : l_ju_ju_kl_k}\end{gathered}\ ] ] in addition we calculate equations for @xmath46 , @xmath47 ( @xmath48 ) , @xmath49 and @xmath50 . in the cavity as a function of @xmath51 and @xmath37 .
the number of structures was chosen to @xmath52 .
the dashed line indicates @xmath53 necessary to obtain cooperative behavior ( valid for @xmath37 small enough to justify setting @xmath54 ) .
( b ) @xmath55 as a function of @xmath51 and @xmath0 ( @xmath56 ) .
the analytical solution gives @xmath57 ( dashed line ) and reliably predicts the onset of cooperative behavior and hence a significant increase in @xmath58.,scaledwidth=50.0% ] approximate analytical solutions can be deduced under certain assumptions and give insight into the main mechanisms @xcite .
we assume that @xmath37 is small enough to justify setting @xmath54 and that @xmath59 . by keeping only the decay term proportional to @xmath24 and the last line in eq . , where we approximate @xmath60
, we obtain the steady - state @xmath61 under the assumption of negligible @xmath37 and by using eq .
it is possible to calculate approximate steady state solutions for all of the involved quantities .
the steady state of the correlation between the structures @xmath62 is of particular interest .
it shows for which parameters the formation of cooperative behavior with the mode is possible @xcite . for a given number of structures
@xmath0 we can calculate the critical temperature @xmath53 that is necessary to meet the condition for cooperative behavior ( @xmath63 ) . at the same time , for a given temperature @xmath51 and the resulting number @xmath64 a certain threshold number @xmath65 of structures is necessary for cooperative behavior .
details can be found in the supplementary material .
the full set of equations is also integrated numerically to determine the exact steady states and to compare them to the approximate solutions .
the parameters @xmath66 ( i.e. fwhm linewidth of @xmath67 ) , @xmath68 , @xmath59 , @xmath69 , @xmath70 , @xmath71 , @xmath72 , @xmath73 , @xmath74 remain fixed for the forthcoming calculations , while @xmath0 , @xmath51 and @xmath37 are varied . the number of photons as a function of @xmath51 and @xmath37 for @xmath52 is shown in fig .
[ fig : laser_photons](a ) .
the occupation of the cavity mode strongly depends on the temperature gradient between the wells .
the approximative analytical treatment for small @xmath37 gives the threshold temperature @xmath53 that signals the onset of cooperative behavior .
for @xmath75 the approximative solutions break down . in fig .
[ fig : laser_photons](b ) we show @xmath76 as a function of @xmath51 and @xmath0 for fixed @xmath56 . here
, the approximative analytical solutions give a reliable result for which parameters to expect a high number of photons in the cavity .
the numerical solutions show that the number of photons in the cavity increases rapidly as soon as we exceed @xmath65 for a given value of @xmath51 .
assume that we have a layer containing @xmath0 structures which initially is hot .
if one side of the layer is cooled by some means , the resulting temperature gradient will lead to a population of the cavity mode .
the rate at which photons are lost from the cavity is @xmath77 .
this suggests that the cavity can serve as a channel for cooling . without the second quantum well , or equivalently with @xmath78 , we find the ratio between the upper level @xmath79 and the lower energy level @xmath80 in the first well to be @xmath81 . in the presence of the second
well , and in the regime where we find a considerable number of photons in the cavity , this ratio is reduced .
figure [ fig : cooling_first_well](a ) shows how the ratio @xmath82 changes with increasing @xmath51 while @xmath83 remains fixed . and @xmath37 .
in ( a ) we choose @xmath83 and vary @xmath51 .
the ratio between the population of @xmath84 and @xmath85 ( solid line ) increases with @xmath51 but remains below the dashed line which gives the ratio in the uncoupled case . in ( b )
the temperature of the hot part @xmath51 is fixed to @xmath86 .
for @xmath87 we again see the reduction of @xmath88.,scaledwidth=40.0% ] with increasing @xmath51 the relative occupation of the upper level @xmath16 increases but remains below the result for the uncoupled case .
the difference starts to show at @xmath89 which is the critical temperature above which we obtain a significant increase in @xmath76 , as can be seen in fig .
[ fig : laser_photons](a ) .
the same comparison can be made for fixed @xmath90 ( leaving @xmath81 fixed ) and varying @xmath37 . while @xmath37 remains low enough to ensure a significant number of photons in the cavity , and hence a considerable dissipation via the cavity mirrors ,
the occupation ratio in the first well is reduced .
this corresponds to an effectively lower temperature of the first well .
we can obtain a large number of photons in the cavity for @xmath91 and an appropriate ratio between @xmath51 and @xmath37 .
the temperature gradient can therefore be seen as an effective pump . to characterize the emission
we calculate the spectrum of the cavity output field and the linewidth . using the quantum regression theorem
we obtain the differential equation for the first order correlation function @xmath92 and @xmath93 , see supplementary material . to find the spectrum we calculate the laplace transform of the coupled equations @xcite .
the steady state values of the involved quantities are obtained numerically . in fig .
[ fig : lw_t1_t2 ] the linewidth ( fwhm ) of the cavity output spectrum is depicted . with the onset of cooperative behavior
the linewidth is reduced dramatically .
similar behavior is found for fixed @xmath37 and varying @xmath51 and @xmath0 . and @xmath37 with @xmath52 ( plotted on a logarithmic scale ) .
the linewidth ( fwhm ) of the empty cavity @xmath94 corresponds to the dark red color ( upper boundary of the scale ) .
the linewidth @xmath95 falls slightly below that limit with increasing @xmath51 until it finally drops off dramatically as the cooperative effects emerge . for small @xmath37
the onset of cooperative behavior is marked by @xmath53.,scaledwidth=50.0% ] the two - well system discussed so far clearly exhibits the essential physical mechanism of heat to coherent light conversion in its purest form . in the following we show that the principle can be extended to multi step excitation processes , where several heat absorbing steps are combined to lead to operation at higher emission frequencies .
mathematically , this can be easily facilitated by incorporating an absorptive step in the second well by choosing @xmath96 , without needing a more complex model . in the dissipative dynamics we simply have to exchange the rates @xmath97 .
physically , this means that in each cycle in this system the absorption of an extra thermal photon or phonon with frequency @xmath98 is necessary .
this implies that we find a minimal temperature below which not enough phonons are present and lasing is impossible . in general one
could of course even think of a combination of thermally assisted and externally driven excitation steps . we will refrain from this here to keep the complexity of the model limited . and @xmath37 .
for @xmath99 we find a sudden increase in @xmath76 , and in ( b ) a corresponding drop in the linewidth of the cavity emission .
the parameters remain unchanged except for @xmath100 and @xmath101.,scaledwidth=50.0% ] let us look at the results now . the temperature range for which @xmath102 becomes significant
is depicted in fig .
[ fig : switch_u_i4_photons_linewidth](a ) .
the dashed line marks the critical temperature necessary for cooperative behavior , which is obtained analytically with similar approximations as described above .
the linewidth of the emission is shown in fig .
[ fig : switch_u_i4_photons_linewidth](b ) . this concept could be incorporated as an extra step in conventional quantum well lasers , which generate intrinsic heat during operation , to extend their range of operation as the absorption of phonons would counteract the heating process .
our theoretical calculations exhibit the possibility to use a spatial temperature gradient in a suitable active medium to induce optical gain and lasing . while not very practicable in the visible regime , a semiconductor heterostructure
can be envisaged to generate narrow bandwidth emission through a resonant cavity mode in the thz regime , where the coherent emission of photons is sustained by the spatial temperature gradient .
the effective extraction of thermal energy from the system in parallel provides for stimulated cooling of the first , hot quantum well . to use a similar idea at higher frequencies without too high temperatures , a multi step excitation process can be envisaged , where some steps could also invoke optical pumping or current injection . in general , introducing an absorptive step well before carrier injection to the upper laser level relies on absorption of thermal quanta and therefore counteracts heating during operation .
this leads to a higher operating temperature limit . while we here only worked with an oversimplified generic model
, the physical principles behind should stay valid for more realistic and thus more complex descriptions .
in addition , transverse spatial temperature gradients or even spatially separated structures connected by conducting wires might provide for alternative configurations at greater length scales .
k.s . was supported by the doc - fforte doctoral program of the aw .
h.r .
acknowledges support from the austrian science fund fwf grant s4013 .
10ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop
_ _ , vol .
( , ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) |
in the present contribution we present experimental results on lagrangian one - particle statistics from an experiment with the ris particle tracking velocimetry ( ptv ) setup .
we focus on small - scale statistics in a turbulent flow : the statistic is analyzed with extended self - similarity ( ess ) @xcite and the results are presented in the spirit of the multifractal model of turbulence @xcite . the use of ess is discussed together with the multifractal model in a finite reynolds number flow like the present .
we have performed a particle tracking velocimetry ( ptv ) experiment in an intermediate reynolds number turbulent flow .
the flow has earlier been reported in @xcite although we use data from a recording with a slightly lower reynolds number .
ptv is an experimental method suitable for obtaining lagrangian statistics in turbulent flows @xcite : lagrangian trajectories of fluid particles in water are obtained by tracking neutrally buoyant particles in space and time .
the flow is generated by eight rotating propellers , which change their rotational direction in fixed intervals in order to suppress a mean flow , placed in the corners of a tank with dimensions @xmath1 ( see fig [ fig : exp ] ) .
the data acquisition system consists of four commercial ccd cameras with a maximum frame rate of @xmath2 at @xmath3 pixels .
the measuring volume covers roughly @xmath4 .
we use polystyrene particles with size @xmath5 and density very close to one .
we follow @xmath6 particles at each time step with a position accuracy of @xmath7 pixels corresponding to less than @xmath8 the stokes number , @xmath9 ( @xmath10 denotes the inertial relaxation time for the particle to the flow while @xmath11 is the kolmogorov time ) is much less than one and the particles can therefore be treated as passive tracers in the flow .
the particles are illuminated by a @xmath12 flash lamp .
the mathematical algorithms for translating two dimensional image coordinates from the four camera chips into a full set of three dimensional trajectories in time involve several crucial steps : fitting gaussian profiles to the 2d images , stereo matching ( line of sight crossings ) with a two media ( water - air ) optical model and construction of 3d trajectories in time by using the kinematic principle of minimum change in acceleration @xcite .
the flow characteristics are presented in table [ table : flow ] . the mean flow is axisymmetric with a significant vertical straining on the largest scales and we did not find any significant differences from the flow reported in @xcite , where properties of the mean flow can be found . here
we look at a sub - volume of the full measuring volume .
only particles which we can observe within a ball of radius @xmath13 is considered and the turbulence characteristics given in table [ table : flow ] are thus only determined from particles inside this ball .
the ball is centered approximately in the center of the tank where the velocity standard deviation @xmath14 has a global minimum . inside the ball
the particles are uniformly distributed . with @xmath15 and a recording frequency at @xmath16 the temporal resolution
is @xmath17 .
.turbulence characteristics : @xmath18 is the mean kinetic energy dissipation , @xmath19 is the kolmogorov length scale with the kinematic viscosity @xmath20 of water .
@xmath21 is the kolmogorov time scale and @xmath22 is the standard deviation of velocity .
the integral length scale is defined as @xmath23 while @xmath24 is the eddy turnover time @xmath25 .
the reynolds number is defined as @xmath26 [ cols="^,^,^,^,^,^,^",options="header " , ] some remarks about ess should be made at this point . in the original paper
@xcite argued , based on experimental evidence of @xmath27 .
as also emphasized in the paper this is not a rigorous result which can be deduced from the navier - stokes equation . by plotting absolute
( defined by positive velocity increments ) structure functions vs. the third order structure function ( eulerian frame ) or the second order structure function ( lagrangian frame ) , an extended scaling range can be observed because undulations in the structure functions are correlated and hence disappear when plotted against each other .
ess is widely used and gives seemingly universal scaling exponents for flows in a wide range of reynolds numbers . as pointed out by @xcite no consensus besides the observed facts exists about the interpretation or even significance of ess .
whether the observed scaling in ess is the signature of hidden physical laws is speculated . in the lagrangian frame
an additional problem arise .
as already mentioned k41 predicts linear dependence of time scale for the lagrangian second order structure function and hence @xmath28 .
this is motivated by the scaling in the eulerian frame and specifically from the _ four - fifth _ law .
a similar _ exact _ result does not exist for the lagrangian structure functions .
so all in all , one could state that _ it is a wonder that it works ! _
the values in table [ table : zeta ] are in excellent agreement with results obtained by @xcite and @xcite .
the values by @xcite are somehow higher and was by @xcite explained as a different choice of inertial range .
the multifractal model of turbulence was introduced by @xcite in the eulerian frame after an early attempt by @xcite who used multifractal measures to characterize the spatial structure of dissipation in turbulence .
the multifractal model is phenomenological and has been able to successfully predict the corrections to k41 similarity scaling for high order moments of spatial velocity increments @xcite .
@xcite discusses multifractals in the lagrangian frame and introduces a bridge to the eulerian framework .
the literature is , however , not very rich on work on lagrangian multifractals , which could have to do with the difficulties in obtaining reliable lagrangian data set more than a animosity against the multifractal model .
work by @xcite have , however , shed light on the issue of multifractals in the lagrangian frame . in the lagrangian multifractal model the flow
is assumed to possess a range of scaling exponents @xmath29 with a certain probability so that @xmath30 for each scaling exponent @xmath31 there is a fractal set with a @xmath31-dependent dimension @xmath32 .
the embedding dimension is one ( @xmath33 ) and hence @xmath34 for all @xmath31 .
the probability @xmath35 of having an exponent @xmath31 at time @xmath36 is therefore proportional to @xmath37 . from a steepest descent argument
one can calculate a relation between the anomalous scaling exponents @xmath38 and the fractal dimension @xmath32 given by @xcite : @xmath39 \label{eq : l1}.\ ] ] if @xmath32 is _ concave _ a legendre transformation gives @xmath40 \label{eq : l2}.\ ] ] in figure [ fig : dh ] we have plotted @xmath32 obtained through eqn .
[ eq : l2 ] .
first we calculated @xmath38 for both integer and non - integer values of @xmath41 between @xmath42 and @xmath43 .
the result is the red curve in the figure [ fig : dh ] . .
the black dots are the result by @xcite . ]
the black dots are the result by @xcite who in a ptv experiment of reynolds number @xmath44 and @xmath45 measured @xmath32 both trough @xmath35 which they manage to measure directly and through eqn .
[ eq : l2 ] as we have done here .
they arrived at the same @xmath32 from both calculations putting confidence in the multifractal model for lagrangian velocity increments .
the agreement between their data and ours is very good . only for @xmath46 in the linear portion of @xmath32 do we observe a discrepancy .
this linear portion of @xmath32 was by @xcite explained in the following way : because the domain of @xmath31 is finite eqn .
[ eq : l1 ] will become a linear function after some @xmath47 .
this linear behavior is also observed in figure [ fig : exponents ] for large @xmath41s .
the transition point @xmath48 happens where @xmath47 minimizes the right hand side of eqn .
[ eq : l1 ] .
for @xmath49 we therefore have that @xmath50 . since only moments of the structure functions of order larger than @xmath51 converge we have @xmath52 and the linear part of the curve is @xmath53 .
@xcite successfully corrected the models by @xcite ( from a theoretical prediction by @xcite ) and @xcite and found a remarkable match .
the discrepancy in figure [ fig : dh ] therefore stems from different estimates of @xmath54 and the uncertainty in measuring it .
@xcite came up with a formula for the connection between @xmath32 and its eulerian counterpart @xmath55 .
the formula is @xmath56 from our database we have calculated the eulerian anomalous scaling exponents from ess structure functions @xmath57 . in ( a ) @xmath58 are plotted from @xmath59 . for
all orders ess seems to work fine . in ( b )
the anomalous scaling exponents @xmath60 are shown .
the values are in perfect agreement with the theoretical model by @xcite .
more interesting is the departure from the k41 prediction which is smaller than in the lagrangian frame .
this is interpret as lagrangian statistics being more intermittent .
just like in the lagrangian frame there is a legendre transformation between @xmath61 and @xmath55 : @xmath62 \label{eq : e1}\ ] ] the only difference from eqn .
[ eq : l2 ] is the embedding dimension which in the eulerian frame is three @xmath63 of order @xmath41 as a function of @xmath64 .
@xmath65 increasing upwards .
( b ) anomalous scaling exponent determined from ( a ) ( red dots ) .
the straight line is the k41 prediction and the green curve is the theoretical model by @xcite . ] from eqn .
[ eq : e1 ] and [ eq : transformation ] we can find @xmath32 from the eulerian anomalous scaling exponent presented in figure [ fig : dheul ] ( b ) .
the comparison is plotted in figure [ fig : dhcomp ] .
again we observe a discrepancy in the linear part of @xmath66 .
whether it comes from the determination of the anomalous scaling exponents from ess or that there is a flaw in eqn .
[ eq : transformation ] we can not say at the moment . a direct measurement of the probability @xmath67 in both the eulerian and lagrangian frame might give more insight into the connection between the two frames . .
the red curve is calculated from lagrangian ess structure functions while the blue is obtained though eulerian structure functions and [ eq : transformation ] . ]
the physical interpretation of the multifractal model is not that easy . in k41 similarity scaling only one scaling exponent is possible , namely @xmath68 and thus @xmath69 .
this is motivated from the fact the navier - stokes ( n - s ) equation is only invariant under one scaling group .
this group is characterized by an exponent @xmath70 obtained by scaling the n - s with the following transformation ( time , position , velocity ) : @xmath71 for @xmath72 .
the solution is @xmath73 . in the limit of infinite reynolds number
the viscosity term in the n - s equation becomes negligible and we find that the n - s equation .
is now invariant to infinitely many exponents @xmath70 .
this is one of the motivations for the multifractal model .
it is , however , not a justification .
another important aspect of the model is the fact that when an eddy breaks up into smaller eddies in the richardson picture the smaller eddies do not cover the same amount of space .
instead they cover only a fraction equal to @xmath74 .
we thus have regions in the flow with large activity and regions with almost calm waters . in the lagrangian frame
this would mean that the individual fluid particles are _ not free _ to move around in all directions .
for example as reported by @xcite and @xcite are particles often trapped by intense vortices .
the large accelerations and velocity increments of these events are therefore of dimension lower than @xmath75 in the eulerian frame and
lower than @xmath76 in the temporal .
this spiral motion of fluid particle around a fluid filament is also the fluid mechanical picture of intermittent events in the model by @xcite : by entraining surrounding fluid kinetic energy fluctuations are effectively dissipated along the axis of the filament .
as emphasized by @xcite the multifractal model does , however , not imply that the trajectories of fluid particles are fractal trajectories themselves .
we have measured lagrangian one - particle statistics and looked at small - scale behavior .
the finiteness of the measuring volume can be used to calculate the mean kinetic energy dissipation @xmath18 in the flow without any further assumptions .
the small scales do seem to be affected by the large - scale inhomogeneities present in our flow .
we do not observe a significant inertial range but by extended self - similarity we are able to extract a quantitative measure of the structure functions of high order . from these
we calculate the lagrangian anomalous scaling exponents and find excellent agreement with already published results . via the multifractal model
we have calculated the lagrangian multifractal dimension spectrum .
the spectrum is similar to the one published by @xcite even though our reynolds number is significantly lower and our mean flow is different .
most importantly we have shown that a high reynolds number is not necessary to obtain results in the lagrangian frame . all experiments and dns
do show the same qualitative features and no clear lagrangian inertial range has been observed .
whether it is because current experiments are performed with too low reynolds number or it simply do not exist future experiments will tell .
the author is grateful to beat lthi , sren ott and jakob mann .
a. arneodo , c. baudet , f. belin , r. benzi , b. castaing , b. chabaud , r. chavarria , s. cilibeto , r. camussi , f. chilla , b. dubrulle , y. gagne , b. hebral , j. herweijer , m. marchand , j. maurer , j. f. muzy , a. naert , a. noullez , j. peinke , f. roux , p. tabeling , w. van de water , and h. willaime .
structure functions in turbulence , in various flow configurations , at reynolds number between @xmath77 and @xmath78 using extended self - similarity . , 34:0 411 , 1996 . | we present lagrangian one - particle statistics from the ris ptv experiment of a turbulent flow .
we estimate the lagrangian kolmogorov constant @xmath0 and find that it is affected by the large scale inhomogeneities of the flow .
the pdf of temporal velocity increments are highly non - gaussian for small times which we interpret as a consequence of intermittency . using extended self - similarity
we manage to quantify the intermittency and find that the deviations from kolmogorov 1941 similarity scaling is larger in the lagrangian framework than in the eulerian . through the multifractal model we calculate the multifractal dimension spectrum . |
the higgs divergent radiative corrections indicate that the standard model ( sm ) is an effective field theory , valid up to a scale @xmath5 .
the excellent agreement between the theory and experiment fixes this scale on a few tev . above this scale a new theory must solve the hierarchy problem .
one possible solution to this problem was recently given by the called little higgs models @xcite .
in these models , the standard model higgs particle is viewed as a pseudo - goldstone boson of a new global symmetry group . in a first stage , this symmetry is spontaneously broken and the higgs is a massless particle . a new `` collective '' symmetry breaking gives a mass to the higgs boson .
the net result is that the new particles and couplings cancel exactly the awful behavior of the standard model divergent diagrams giving a light mass to the higgs fields .
recently , several proposals for `` little higgs '' models have been made .
their main difference are the choice of the new global symmetry .
one explicit model @xcite , with fewer number of new parameters as possible , was named as the `` littlest higgs model '' ( here after lhm ) .
it is constructed using an @xmath6 coset : a gauged @xmath7 is broken to its subgroup @xmath8 . a detailed study of this model was done in reference @xcite and some more general reviews are given in @xcite .
the earlier versions of the lhm presented some difficulties . from the analysis of the electroweak precision constraints it was shown @xcitethat a fine tuning was necessary in order to get a higgs mass of @xmath9 gev .
the main problem could be related to the large top yukawa coupling that generates instabilities in the higgs potential .
some modifications were introduced such as a new single u(1 ) @xcite and t - parity @xcite .
the important point in these extentions is that the scale parameter that fixes the lhm could be as low as @xmath10 gev instead of @xmath11 tev .
this point re - opened the possibility of testing the lhm at the lhc energies .
many proposed models predict the existence of an extra heavy neutral gauge boson , @xmath12 . these include : 3 - 3 - 1 models @xcite , little higgs model @xcite , left - right symmetric models @xcite , supersymmetric breaking with an extra non - anomalous u(1 ) @xcite , superstring inspired e@xmath13 model @xcite and models with extra dimensions as kaluza - klein excitations of neutral gauge bosons @xcite .
the experimental verification of these models , and alternatives like supersymmetric models , could be done by the small quantum corrections to the sm predictions and/or by the discovery of new particles .
this will be the main task of the high energy and luminosity accelerators lhc and ilc .
since there are many models with new interactions and particles , it will be necessary to have very clear signatures to distinguish the available models . from the experimental point of view
a fundamental question is the nature and properties of the lightest particle in extended models . in the lhm
this particle is a new neutral gauge boson , named @xmath14 .
a simple and unique property of this model is the presence of a second , heavier , new neutral gauge boson , named @xmath15 with correlated masses .
these new neutral heavy gauge bosons ( named in general here as @xmath16 ) could be found in the next generation of high energy colliders : hadron @xmath17 hadron @xmath18and @xmath19 ( with @xmath20 ) .
these topics have been studied by many authors during the last years @xcite .
it is expected that at the lhc energies the new neutral gauge bosons will give detectable signals or new mass limits for the @xmath16 and that at future high energy electron - positron colliders it will be possible to study its properties , in details .
the center - of - mass energy of the future colliders will be set to the highest possible values , consistent with technical feasibility and cost limits .
however it is also possible that the masses of the new particles will be below the initially designed collider energy .
another fundamental point is that the little higgs models , as well as practically all other extended model with a new neutral gauge boson , have at least two unknown parameters : the new mass value and some mixing parameter(s ) in the coupling with standard matter .
the optimistic view that a new neutral current experimental signal will be seen at the lhc and more precisely studied at a new lepton collider will probably show that many extended model could account for this signal .
but the lhm has another fundamental feature : a second , heavier , new neutral gauge boson . as cross sections at the resonant energies
have the highest possible values it will be necessary to recalibrate the whole collider in order to fix its cm energy at the new resonant mass .
although physically well motivated , this procedure could be time - consuming and costly .
it is important to have simple alternative signatures in order to show model differences already in the initial collider working period .
earlier works on the lhm have started from the possibility of a first experimental signature for new gauge interactions at the lhc and then the leptonic colliders were considered at a fixed center of mass energy equal to the new boson mass values @xcite .
the main point proposed in this paper is that it is not necessary to wait until the lepton collider energy could be fixed at the mass of the new neutral gauge boson in order to get its properties and exclude some models .
we will consider new bounds and model differences of new neutral gauge bosons for different @xmath21 values , taking @xmath22 , both for polarized and unpolarized beams .
the other difference with previous works @xcite is that we will consider as a signature for new neutral gauge bosons the process @xmath23 , where the final state photon is a hard photon with @xmath24 gev @xcite . in this paper
we will discuss these two main signatures for electron - positron colliders that can point to model differences : asymmetries in the production of fermion pairs and the production of a single hard photon associated with new heavy neutral gauge bosons .
total cross sections will be shown to present detectable effects .
the fermion pair production allows one to find new limits that can be model dependent for @xmath25 and the hard photon production allows one to study the region for @xmath26 .
we concentrate our attention first on the lightest neutral gauge boson of the lhm - @xmath27 .
the present experimental limit is @xmath28 gev obtained by @xcite and is somewhat model dependent .
the other neutral gauge boson @xmath29 is heavier , with a mass of the order @xmath30 . in section 2
we briefly review the models considered in this paper . in section 3
we present our results on the asymmetries .
the new bounds on the lhm parameters are displayed in section 4 , and a new signature for the @xmath14 and @xmath31 bosons in section 5 .
finally , in section 6 , we give our final conclusions .
in order to compare the lhm predictions with other models we will employ the canonical @xmath32 superstring inspired @xmath33 models @xcite , symmetric left - right model ( slrm ) and mirror left - right model ( mlrm ) @xcite . as our main interest is the detection of new gauge bosons
, we will consider only their coupling to ordinary matter .
new particle states must also be coupled to new gauge bosons but this will introduce new parameters and will make the estimates more model dependent in the order hand this will demand also a more sophisticated experimental analysis . the general form of the interaction lagrangian involving only ordinary fermions and extra neutral gauge bosons is given by @xmath34 where @xmath35 and @xmath36 are respectively the vector and vector - axial couplings .
the @xmath37 and @xmath38 couplings to the ordinary matter are the same as the sm ones . for the lhm , according to reference @xcite , the new neutral gauge bosons @xmath14 and @xmath15 couple with ordinary matter with mixing angles @xmath39 and @xmath40 respectively .
the couplings are shown in tables [ tab1 ] and [ tab2 ] , where @xmath41 and @xmath42 are the constant couplings group , @xmath43 and @xmath44 are hypercharges and @xmath45 , are combinations of the coupling parameters @xmath46 and @xmath47 . .
couplings between @xmath14 and fermions in lhm , with @xmath48 where @xmath39 is the mixing angle . [ cols="^,^,^ " , ]
the angular asymmetry is an interesting variable because it is strongly dependent on the couplings @xmath35 and @xmath36 .
the leptonic asymmetries for the process @xmath49 presented in @xcite were calculated for @xmath50 where @xmath1 is the lightest new neutral boson mass . in this paper , we consider the same channel but for new gauge boson mass regions below and above the collider energy . in all calculations ,
each final lepton was required to be detected within the azimuthal angle range @xmath51 , where @xmath52 is the angle of either of the final fermions with respect to the direction of the electron beam .
this cut corresponds roughly to the detector limitations . for the lhm ,
the total width @xmath53 was calculated considering the main contributions from pairs of ordinary charged and neutral fermions and @xmath54-@xmath38-@xmath55 couplings . for the @xmath56 width
we have included these channels and the relevant contribution from @xmath57 channel , which is negligible for @xmath14 .
the mass relation between @xmath14 and @xmath15 is approximately given by the relation @xmath58 .
as a consequence , in the energy region from @xmath59 gev to @xmath60 tev the processes will be dominated by @xmath38 and @xmath14 exchange , if we assume @xmath61 gev .
typical values for the widths , for a choice @xmath62 , @xmath63 , give us @xmath64 gev for @xmath65 tev and @xmath66 gev for @xmath67 gev , in accordance with the results in reference @xcite . a first result involving lhm parameters is the total cross section for @xmath68 as a function of the @xmath21 , for @xmath69 gev and @xmath70 gev . the curves are shown in figure [ fig1 ] . as expected , near the resonant region
there is a peak related to the extra gauge boson exchange .
we have considered the mixing angles @xmath40 and @xmath71 in the ranges of 0 - 0.5 and 0.62 - 0.73 , respectively , as allowed by the precision electroweak constrains on the parameter space of the lhm @xcite . for @xmath69 and @xmath72 gev for lhm ( @xmath73 and @xmath74).,scaledwidth=50.0% ] a first difference between the theoretical models
can be found in the forward - backward asymmetries ( @xmath75 ) in @xmath76 , with @xmath77 and @xmath78 .
these fermion couplings to @xmath14 allow a clear model separation from other extended models .
we have performed the calculation with the comphep package @xcite , in which the previously mentioned models were implemented .
the curves for @xmath75 are displayed in figures [ fig2],[fig3 ] and [ fig4 ] for the models cited above : lhm with a choice of the mixing parameters @xmath79 , @xmath80 ; superstring inspired models @xmath81 and symmetric and mirror left - right models as a function of @xmath82 ( and @xmath1 ) for @xmath83 gev . near the resonant region ( @xmath83 gev ) some models are more sensitive than the lhm , presenting sizeable deviations from the sm value @xmath84 .
the same behavior occurs for @xmath85 tev . for larger values of the mass of the new gauge boson
, the combined analysis of @xmath86 , @xmath79 and @xmath78 pair production might help to establish the best model underlying these deviations from the sm predictions .
( @xmath1 ) for @xmath87 gev in @xmath88 for some models.,width=283,height=171 ] ( @xmath1 ) for @xmath87 gev in @xmath89 for some models.,width=283,height=170 ] ( @xmath1 ) for @xmath87 gev in @xmath90 for some models.,width=283,height=170 ] another variable that shows differences between models is the left - right asymmetry ( @xmath91 ) .
the degrees of polarization of the electron and positron beams were taken to be @xmath92 and @xmath93 respectively .
we present in figures [ fig5 ] for @xmath94 , [ fig6 ] for @xmath95 and [ fig7 ] for @xmath96 , the left - right asymmetry for different models as a function of @xmath82 for @xmath83 gev . again
the @xmath91 for the lhm has a small deviation from the sm value .
the @xmath91 for all models has a smaller deviation from the sm than the @xmath75 for the corresponding models .
the polarized beams will not improve the model distinction with respect to the unpolarized beams . versus @xmath82 ( @xmath1 ) for @xmath83 gev for some models.,scaledwidth=50.0% ] versus @xmath82 ( @xmath1 ) for @xmath83 gev for some models.,scaledwidth=50.0% ] versus @xmath82 ( @xmath1 ) for @xmath83 gev for some models.,scaledwidth=40.0% ]
bounds on the lhm were previously obtained in @xcite at the energy peak of the new neutral gauge boson production , using a combination of physical observables at that energy .
deviations from the sm predictions were shown to be within these bounds . in this paper
we present new bounds on the mass @xmath1 of the new neutral gauge boson @xmath14 as a function of @xmath97 .
these new bounds are derived from the angular distribution of the final - state fermion , by means of a @xmath98 test .
assuming that the experimental data in the fermion - pair production will be described by the sm predictions , we defined a @xmath98 estimator @xmath99 where @xmath100 is the number of sm events collected in the @xmath101 bin , @xmath102 is the number of events in the @xmath101 bin as predicted by the lhm , and @xmath103 the corresponding total error , which combines in quadrature the poisson - distributed statistical error with the systematic error .
we took @xmath104 as the systematic error in our calculation .
we considered the muon , charm and bottom detection efficiencies equal to @xmath105 , @xmath93 and @xmath106 respectively .
we estimated upper bounds for @xmath1 with @xmath105 c.l .
, fixing @xmath73 and using the mixing angle @xmath39 as a free parameter .
figure [ fig8 ] shows the bounds on @xmath1 , for @xmath83 gev and an integrated luminosity @xmath107 fb@xmath108 .
the @xmath105 c.l .
bounds for @xmath85 tev and @xmath109 fb@xmath108 are displayed in figure [ fig9 ] . in both cases
the muon channel was found to be more restrictive than the hadron channels .
bounds ( @xmath105 c.l . ) as a function of @xmath110 in @xmath111 , where @xmath112 and @xmath78 for @xmath83 gev for lhm.,scaledwidth=50.0% ] bounds ( @xmath105 c.l . ) as a function of @xmath110 in @xmath111 , where @xmath112 and @xmath78 for @xmath85 tev for lhm.,scaledwidth=50.0% ] in figure [ fig10 ] we display the resulting limits on @xmath1 for @xmath83 gev , but considering longitudinally polarized beams .
the degrees of polarization of the electron and positron beams were taken to be @xmath92 and @xmath93 respectively .
the limits obtained in this case are less restrictive than the corresponding bounds derived from unpolarized angular distributions , and are essentially the same for all three final states . bounds ( @xmath105 c.l . ) as a function of @xmath113 in @xmath111 , where @xmath114 and @xmath78 for @xmath83 gev for lhm , considering polarized beams.,scaledwidth=50.0% ]
a second point that can show interesting properties of a new fundamental interaction mechanism is the associated production of a general @xmath16 and a hard photon in the process @xmath115 @xcite .
this channel is suitable when @xmath116 .
the main advantage of this channel is to show that one can study the new @xmath12 properties before all the redefinition of the collider energy at the new neutral gauge boson mass .
the total cross section for this channel is smaller than the direct resonant cross section but can give a statistically meaningful number of events , since the expected luminosity for the ilc will be larger than @xmath117 fb@xmath108 .
another important advantage is that one does not need to reconstruct the heavy neutral boson mass from its decay products .
the direct @xmath12 decay is important , but the reconstruction efficiency will reduce the total number of events , as discussed in @xcite .
this reconstruction can be more complicated when these new bosons decay into unstable particles , like hadrons .
the hard photon emission proposed in this paper is very different from the usual multiphoton emission accompanying @xmath12 production .
the well known logarithm corrections @xcite were studied more recently @xcite and imply changes at the @xmath12 pole . but these corrections are different from the kinematics of two body final state in the associated hard photon production @xcite .
a very simple consequence of the conservation of energy and momentum , is that the final high energy hard photon has a fixed energy given by @xmath118 where @xmath119 and @xmath120 are the uncertainties in the photon energy and @xmath82
. the study of the hard photon energy distribution will give the same information as the direct @xmath16 decays , but in a simple and direct way since it is not necessary to reconstruct the @xmath12 subproducts .
the resulting photon energy distribution contains important information about the @xmath14 and @xmath15 bosons and is shown in figures [ fig11 ] and [ fig12 ] , for the process @xmath121 .
similar results would be obtained for @xmath122 . for @xmath123 tev if @xmath124 gev , the photon energy distribution @xmath125 shows two peaks , coming from @xmath14 and @xmath15 , when considering the lhm . for @xmath126 gev
only the peak associated to @xmath14 will show up .
the figures were obtained using a cut @xmath24 gev on the hard photon energy , @xmath127 gev and @xmath128 , where @xmath129 and i = @xmath130 .
the @xmath24 gev cut eliminates the contributions from the @xmath37 s soft emissions .
we notice the strong mixing - angle dependence on the magnitude of the photon peaks in figure [ fig11 ] .
the peak associated to @xmath15 will depend on @xmath131 and the peak associated to @xmath14 will vary with @xmath132 .
the @xmath133 and @xmath134 models have a much larger value for the total neutral gauge boson width and this fact makes the hard photon energy much broader as shown in figure [ fig12 ] .
the @xmath135 distribution allows us to find the @xmath16 mass and to distinguish the models , in connection with eq .
( 3 ) .
showing two peaks associated to @xmath136 gev ( third peak ) and @xmath137 gev ( first peak ) with @xmath138 tev for lhm ( @xmath73 and @xmath74 ) , and only one ( central ) peak when @xmath139 gev .,width=264,height=188 ] with @xmath138 tev for @xmath140 , @xmath141 , @xmath142 and for lhm ( @xmath73 and @xmath74).,width=264,height=188 ]
we have shown that @xmath75 is better than @xmath91 in @xmath143 , where @xmath144 , @xmath78 and @xmath79 , for @xmath145 , and can distinguish the lhm from other proposed models in a wide @xmath12 mass range .
a @xmath98 analysis on the angular distribution of the final fermions for the same channels allows us to obtain a more restrictive @xmath12 mass limits as a function of @xmath132 .
the above calculations were done considering polarized and unpolarized beams . using the angular distribution it is possible to get other mass limits than using simply the total cross - sections as done by @xcite . for @xmath146 ,
the hard photon energy distribution in @xmath147 can indicate the theoretical origin of new possible heavy neutral gauge bosons independent of how these new bosons decay .
this alternative signature for @xmath12 production at the new electron - positron colliders could allow us to study its properties , at a fixed collider energy . if @xmath148 only one peak shows up in the hard photon energy distribution , and the models can be distinguished using the width of this peak . if @xmath149 two peaks appear in the photon energy distribution , and all the models will be excluded , except the lhm . in order to detect a new gauge boson and
understand its properties , the center - of - mass energy of the future high - energy electron - positron experiments will have to be set to the mass of this new boson .
the main point presented in our work is to show that we can establish the same properties of the possible new gauge boson when @xmath150 since the first periods of collider data taking , as shown in figures 11 and 12 .
abc n. arkani - hamed , a. g. cohen and h. georgi , phys .
b * 513 * , 232 ( 2001 ) .
[ arxiv : hep - ph/0105239 ] . n. arkani - hamed , a. g. cohen , e. katz and a. e. nelson , jhep * 0207 * , 034 ( 2002 ) .
[ arxiv : hep - ph/0206021 ] .
t. han , h. e. logan , b. mcelrath and l. t. wang , phys .
d * 67 * , 095004 ( 2003 ) .
[ arxiv : hep - ph/0301040 ] .
m. schmaltz and d. tucker - smith , [ arxiv : hep - ph/0502182 ] .
m. perelstein , [ arxiv : hep - ph/0512128 ] . c. csaki , j. hubisz , g. d. kribs , p. meade and j. terning , phys .
d * 68 * , 035009 ( 2003 ) ; t. gregoire , d. r. smith and j. g. wacker , phys . rev .
d * 69 * , 115008 ( 2004 ) .
m. perelstein , m. e. peskin and a. pierce , phys .
d * 69 * , 075002 ( 2004 ) .
[ arxiv : hep - ph/0310039 ] .
i. low , jhep * 0410 * , 067 ( 2004 ) .
[ arxiv : hep - ph/0409025 ] .
j. hubisz and p. meade , phys .
d * 71 * , 035016 ( 2005 ) .
[ arxiv : hep - ph/0411264 ] .
j. hubisz , p. meade , a. noble and m. perelstein , jhep * 0601 * , 135 ( 2006 ) .
[ arxiv : hep - ph/0506042 ] .
f. pisano and v. pleitez , phys .
d * 46 * , 410 ( 1992 ) .
p. h. frampton , phys .
* 69 * , 2889 ( 1992 ) .
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 ) . an extensive list of references can be found in r. n. mohapatra and p. b. pal , `` massive neutrinos in physics and astrophysics '' , world scientific , singapore , 1998 . m. cvetic , d. a. demir , j. r. espinosa , l. everett and p. langacker , phys . rev . * 56 * , 2861 ( 1997 ) . j. l. hewett , t. g. rizzo , phys . rep . *
183 * , 193 ( 1989 ) .
t. g. rizzo , e - conference c040802 , l013 ( 2004 ) .
american linear collider group , [ arxiv : hep - ex/0106055 v1 ] .
physics at an @xmath0 linear collider , tesla technical design report , part iii , eds .
r. -d .
heuer , d. miller , f. richard , p. m. zerwas , march 2001 .
f. abe et al . , phys .
lett . * 79 * , 2192 ( 1997 ) ; s. obachi et al .
b * 385 * , 471 ( 1996 ) .
j. a. conley , j. hewett and m. p. le , phys .
d * 72 * , 115014 ( 2005 ) .
[ arxiv : hep - ph/0507198 ] .
s. godfrey , p. kalyniak and a. tomkins , [ arxiv : hep - ph/0511335 ] .
f. m. almeida , y. a. coutinho , j. a. m. simes , a. j. ramalho , s. wulck and m. a. b. do vale , eur .
j. c * 44 * ( 2005 ) 127 .
[ arxiv : hep - ph/0412395 ] .
particle data group , phys .
b * 592 * , ( 2004 ) 1 .
chong - xing hue , wei wang , feng zhang , nucl .
b * 716 * , ( 2005 ) 199 .
[ arxiv : hep - ph/0409066 ] .
gi - chol cho and aya omote , phys .
d * 70 * , ( 2004 ) 057701 .
a. pukhov , e. boos , m. dubinin , v. edneral , v. ilyin , d. kovalenko , a. kryukov , v. savrin , s. shichanin and a. semenov , `` comphep''- a package for evaluation of feynman diagrams and integration over multi - particle phase space .
preprint inp msu 98 - 41/542 [ arxiv : hep - ph/9908288 ] o. nicrosini and l. trentadue , phys .
b * 196 * , ( 1987 ) 551 .
a. freitas , phys .
d * 70 * , ( 2004 ) 015008 .
x. wang , q. zeng , z. jin and s. liu , [ arxiv : hep - ph/0702064 ] . | there is a recent proposal identifying the higgs particle of the standard model as a pseudo nambu - goldstone boson .
this new broken symmetry introduces new particles and new interactions . among these new interactions a central role to get a new physics is played by two new heavy neutral gauge bosons .
we have studied the two new neutral currents in the littlest higgs model and compared with other extended models . for high energy @xmath0 colliders we present a clear signature for these two new neutral gauge bosons that can indicate the theoretical origin of these particles .
previous analysis by other authors were done at collider energies equal to the new gauge boson mass @xmath1 . in this paper
we show that asymmetries in fermion antifermion production can display model differences in the case @xmath2 .
for @xmath3 we show that the hard photon energy distribution in @xmath4 can present a model dependence . for higher energies ,
the hard photon energy distribution can be a clear signature for both new neutral gauge bosons .
new bounds for the new neutral gauge boson masses are also presented . |
the paradigm of cosmic inflation @xcite ( for recent reviews see e.g. @xcite ) has proven very successful in resolving the flatness and horizon problems of the early universe and in explaining the absence of relics from early phase transitions . however , the connection to particle physics is still unclear .
one possibility to establish such a connection is provided by sneutrino hybrid inflation @xcite , where the superpartner of one of the right - handed neutrinos involved in the seesaw mechanism @xcite plays the role of the inflaton field . in sneutrino hybrid inflation , a large vacuum energy density is present which drives inflation and the sneutrino direction in field space has an almost flat potential suitable for slow - roll inflation . the right - handed ( s)neutrinos obtain their large masses after the `` waterfall '' phase transition which ends hybrid inflation . inflation in this scenario is closely linked to the physics generating the small neutrino masses via the seesaw mechanism . another attractive connection between the seesaw mechanism and early universe cosmology is the possibility of generating the observed baryon asymmetry of the universe via the out - of - equilibrium decays of the right - handed ( s)neutrinos in leptogenesis @xcite ( for recent reviews see @xcite ) . for calculating the produced baryon asymmetry , the knowledge of the phase of ( p)reheating after inflation is in general mandatory , since it may lead to the nonthermal production of right - handed ( s)neutrinos and since it determines the reheat temperature , which in turn governs the possibility of thermal ( s)neutrino production . in most inflation models
the nonthermal ( s)neutrino production must arise from the decays of the inflaton field . in sneutrino hybrid inflation , on the other hand , the inflaton itself is a right - handed sneutrino , which means that this intermediate step is skipped and the sneutrino inflaton field after inflation may directly dominate the universe and , when it decays , most efficiently produce the baryon asymmetry and reheat the universe .
in previous works , leptogenesis after sneutrino inflation has been studied in the context of chaotic sneutrino inflation @xcite which however requires a quite heavy sneutrino with a mass of about @xmath0 gev and correspondingly very small yukawa couplings in order to realize a low reheat temperature as suggested by gravitino and similar constraints in supersymmetric cosmology .
furthermore , chaotic sneutrino inflation with a quadratic potential for the inflaton gives rise to a comparatively large tensor - to - scalar ratio of @xmath1 . on the other hand , sneutrino hybrid inflation @xcite , as typical for hybrid - type inflation scenarios @xcite , predicts a much smaller ratio @xmath2 and is thus clearly distinguishable from chaotic sneutrino inflation by future observations ( e.g. by the planck satellite ) .
recently , it has been shown that sneutrino hybrid inflation @xcite belongs to a wider class of hybrid - like inflation models , dubbed `` tribrid inflation '' in @xcite , which are very suitable for being embedded into supergravity ( sugra ) theories with the sugra @xmath3-problem solved by either a shift symmetry @xcite or a heisenberg symmetry @xcite in the khler potential .
while the sneutrino was a gauge singlet in @xcite , it has been demonstrated in @xcite that it may be embedded into a grand unified theory ( gut ) representation , e.g. into a * 16 * -plet of so(10 ) , establishing a possible link between sneutrino hybrid inflation and left - right symmetric guts .
nonthermal leptogenesis after sneutrino hybrid inflation , on the other hand , was only briefly discussed in @xcite for an example set of model parameters . in this paper
, we therefore investigate in detail the conditions under which inflation , with primordial perturbations in accordance with the latest wmap results , as well as successful nonthermal leptogenesis can be realized simultaneously within the sneutrino hybrid inflation scenario .
we point out which requirements successful inflation and leptogenesis impose on the seesaw parameters , i.e. on the yukawa couplings and the mass of the right - handed ( s)neutrino , and derive the predictions for the cmb observables in terms of the right - handed ( s)neutrino mass and the other relevant model parameters .
our results are meant as a guideline for the construction of explicit particle physics models incorporating sneutrino hybrid inflation and baryogenesis via nonthermal leptogenesis .
the paper is organized as follows : in section [ framework ] we introduce the sneutrino hybrid inflation scenario in a simple setup .
section [ inflation ] is dedicated to the inflationary phase and the predictions for the cmb observables . in section [ leptogenesis ]
we discusses the reheating of the universe after inflation and the production of the baryon asymmetry of the universe via nonthermal leptogenesis .
we conclude in section [ infl_lepto ] by combining our results from inflation and leptogenesis to highlight the preferred ranges of the model parameters .
we will discuss sneutrino hybrid inflation and subsequent baryogenesis via nonthermal leptogenesis in an extension of the minimal supersymmetric standard model ( mssm ) with conserved r - parity , where three additional right - handed ( s)neutrinos acquire large masses after the `` waterfall '' phase transition ending inflation . the superpotential defining our framework
is given by @xmath4 where the @xmath5 ( the index @xmath6 denotes the different generations ) are gauge singlet superfields describing the heavy right - handed ( s)neutrinos and where the reduced planck scale is given by @xmath7 .
the canonically normalized imaginary parts - problem as was demonstrated in @xcite . ]
@xmath8 of the respective scalar components are inflaton candidates as will be described below . @xmath9 and @xmath10 are su(2)@xmath11-doublet superfields which contain the standard model leptons and up - type higgs , respectively .
the yukawa coupling term of @xmath12 with the higgs and lepton doublet , i.e. the second term in eq . , allows to identify @xmath12 with the right - handed neutrino superfield .
@xmath13 and @xmath14 are two additional gauge singlet superfields . here
the canonically normalized real part @xmath15 of the scalar component of @xmath13 is the so - called `` waterfall '' field responsible for ending inflation .
the f - term of @xmath14 , the so - called `` driving superfield '' , provides the large vacuum energy density that drives inflation .
the scalar component of @xmath14 is fixed at zero during inflation by sugra corrections ( cf .
section [ sugra ] ) and does not affect the inflationary dynamics . furthermore , we assume @xmath16 and @xmath17 to be real coupling parameters for simplicity . the form of the superpotential eq .
is motivated as follows : the latter two terms generate the scalar potential suitable for inflation . in the false vacuum with large values of @xmath8 and @xmath15 stabilized at zero , the large vacuum energy @xmath18 drives the quasi - exponential growth of the scale factor in inflation .
once the slow - rolling fields @xmath8 fall below a critical value , the negative contribution to the squared mass of @xmath15 from the term @xmath19 starts dominating over the positive contribution from the terms @xmath20 .
therefore , @xmath15 becomes tachyonic which triggers the `` waterfall '' ending inflation as @xmath15 acquires a non - zero vacuum expectation value ( vev ) .
after inflation , close to the global minimum of the potential where @xmath21 and @xmath22 and where the large vacuum energy contribution vanishes , the fields @xmath8 and @xmath15 perform damped oscillations accounting for a matter dominated universe .
the field which decays last and finally dominates the universe is generically the right - handed sneutrino with the smallest mass and smallest yukawa couplings , since it only decays via the second term in eq .
proportional to @xmath23 .
this decay reheats the universe which thus enters its radiation dominated epoch . for illustration ,
we have plotted a typical scalar potential resulting from the scenario described above in fig .
[ hybridpotential ] . for further details see e.g. ref .
@xcite . typical f - term scalar potential resulting from the model of eq . for the scalar components of @xmath24 set to zero . for the plot ,
we have used example parameters @xmath25 and @xmath26 . ] the first three terms in eq . describe the mssm with masses for the additional right - handed neutrinos generated after inflation . in particular the term
@xmath27 generates mass terms for the heavy ( s)neutrinos as @xmath15 acquires its non - zero vev .
the vev of @xmath15 in the true minimum is governed by the fourth term in eq . .
in a realistic scenario , we would expect inflation to end by a phase transition , i.e. the @xmath15 field to be a non - singlet under some symmetry group .
might be replaced by a different combination of fields , i.e. by @xmath28 , in order to form a singlet under the symmetry group .
the symmetry could , e.g. , be a unifying gauge symmetry or a family symmetry . ] a realization within the context of guts can be found in ref .
@xcite , and the idea of relating the `` waterfall '' of hybrid inflation to the breaking of a family symmetry was discussed in ref .
@xcite . for simplicity ,
we keep @xmath13 a gauge singlet here .
finally , the ellipsis represent possible higher dimensional operators .
we note that a superpotential of the form given in eq .
can be realized in an explicit model with discrete symmetries , as has been shown in @xcite .
these discrete symmetries distinguish between the gauge singlet fields of eq . .
the parameters which appear in eq . , and which govern our framework , can be understood as follows : * the * phase transition scale * @xmath29 is the vev of the scalar component of the @xmath13 superfield after the phase transition ending inflation and is the mass scale relevant for inflation . *
the parameter @xmath30 determines the * seesaw scale * which corresponds to the mass of the lightest right - handed neutrino @xmath31 in the true vacuum of the theory . * the * vacuum energy parameter * @xmath17 fixes the the vacuum energy density with regard to the phase transition scale . *
the * effective yukawa coupling * @xmath32 is linked to the light neutrino masses .
we will be working in a sugra framework with sugra corrections stabilizing the scalar components of @xmath14 , @xmath9 and @xmath10 during inflation and with a symmetry in the khler potential that guarantees tree - level flat @xmath8 directions .
this solves the @xmath3-problem of sugra inflation .
the details of such a sugra framework are discussed in section [ sugra ] . however , to illustrate the underlying physics more clearly we first focus on a global supersymmetry ( susy ) model and take the features mentioned above for granted . in order to produce the cp - violation necessary for leptogenesis we work with three ( s)neutrino generations .
assuming that the right - handed neutrinos are strongly hierarchical , i.e. one of them is significantly lighter than the other two , the scalar components of the latter superfields can be stabilized at their minima before the final 60 e - folds of inflation begin .
thus the time evolution of the lightest sneutrino controls the relevant slow - roll dynamics and it can therefore be identified as the inflaton . on the other hand ,
the outcome of leptogenesis is governed by the sneutrino with the smallest decay rate .
this implies a comparatively small mass and small yukawa couplings . in the following ,
we shall concentrate on the case where the lightest sneutrino drives both inflation and leptogenesis .
hence , the three generation model can be simplified to an effective one generation model in the right - handed neutrino sector , with the only remnant of the other two generations being a non - vanishing cp - asymmetry necessary for leptogenesis .
we can thus concentrate on @xmath33 in eqs .
and we denote the relevant inflaton direction by @xmath34 and the respective coupling constant by @xmath35 .
based on the framework described in the previous section , we now have a closer look at the inflationary dynamics in our model .
furthermore , we derive restrictions on the model parameters from the requirement of successful inflation and the latest observational data .
we start with a short introduction to slow - roll inflation and then discuss a realization of the model of section [ framework ] in a globally supersymmetric context .
we then refine this discussion by including sugra effects and close the section by listing the inflationary predictions from our model and comparing them to the latest observational data . a common way to realize inflation is the so - called slow - roll paradigm , where a classical scalar field with a strong dominance of its potential energy over its kinetic energy @xmath36 drives the accelerated expansion of the universe .
at the same time , the quantum fluctuations of the inflaton field can account for the metric perturbations which give rise to the small scale cmb anisotropies .
inflation ends when the slow - roll conditions are violated , i.e. when the slow - roll parameters parameterizing the scalar potential and its derivatives @xmath37 become of order one .
here , a prime denotes derivative w.r.t . the inflaton field @xmath38 .
in the slow - roll approximation , when @xmath39 , @xmath40 and @xmath41 , the equation of motion of a homogeneous ( classical ) scalar field @xmath42 simplifies to @xmath43 here @xmath44 denotes the hubble expansion parameter .
models of inflation typically predict the power spectra of the gauge invariant scalar and tensor perturbations at the time when the relevant fluctuations exited the horizon , roughly @xmath45 e - folds before the end of inflation .
the amplitude of the scalar perturbations @xmath46 , the scalar spectral index @xmath47 , the running of the scalar spectral index @xmath48 , the tensor - to - scalar ratio @xmath49 and the tensor spectral index @xmath50 can be estimated in terms of the potential and the slow - roll parameters @xcite as @xmath51 where these expressions have to be evaluated at the field value @xmath52 when the relevant scales leave the horizon .
this value can be computed from eq . .
in order to test our model against observations , we compare the predictions from eqs . to the experimental data obtained from the 7 year wmap survey combined with measurements of the baryon acoustic oscillations ( bao ) @xcite and measurements of the present value of the hubble parameter @xmath53
@xcite using the six parameter @xmath54cdm fit @xcite , which are given by @xmath55 in the model described in section [ framework ] , we assumed a symmetry in the khler potential guaranteeing a flat @xmath38 direction ( imaginary direction of the scalar component of @xmath56 ) at tree - level ( see also section [ sugra ] ) .
thus loop corrections must be taken into account and these can indeed generate a small slope as required for slow - roll inflation . according to @xcite , the coleman - weinberg one - loop effective potential is given by @xmath57 \,,\ ] ] with @xmath58 denoting the mass matrix of the theory and @xmath59 a renormalization scale .
the @xmath38-dependent bosonic and fermionic mass terms generating a slope for the inflaton via the loop potential can be calculated from the scalar f - term potential @xmath60 and the fermionic mass matrix @xmath61 @xmath62 denote the superfields of the theory , @xmath63 the respective scalar components .
the relevant ( i.e. @xmath38-dependent ) contributions to the loop potential are give by the @xmath64 and @xmath10 mass terms @xmath65 with @xmath66 note that the @xmath67 and @xmath10 terms in the supertrace vanish since the degeneracy in the respective fermionic and bosonic masses leads to a cancellation of these contributions .
embedding this model in sugra provides the necessary stabilization of the scalar components of the @xmath67 and @xmath10 superfields during inflation and removes this degeneracy .
however in the parameter range of interest , the contribution of the @xmath67 and @xmath10 terms to the loop potential turn out to be negligible ( see also section [ sugra ] ) . in the following ,
we fix the renormalization scale to @xmath68 , which is the order of magnitude of the susy breaking scale . in our model ,
inflation ends when the @xmath15-field destabilizes at the critical value @xmath69 characterized by @xmath70 and thus @xmath71 , @xmath72 we can now determine the observables describing the cmb fluctuations given by eqs .
, thus obtaining expressions depending on the phase transition scale @xmath29 , the seesaw scale @xmath73 and the vacuum energy parameter @xmath17 .
with @xmath74 as defined above , a taylor expansion in @xmath75 ( with @xmath76 because @xmath77 during inflation ) yields @xmath78 inserting this into the equation of motion eq . , with @xmath79
approximately constant , gives the value for @xmath38 at @xmath80 e - folds before the end of inflation : @xmath81 with this , the inflationary predictions of eqs .
are given by @xmath82 we next consider a possible embedding of our model in sugra .
we focus on a khler potential with the @xmath3-problem @xcite resolved by a shift symmetry @xcite in the inflaton direction @xmath83 the khler potential can be seen as a general expansion in the superfields of the theory with the additional feature of a shift symmetry which guarantees tree - level flat directions for the imaginary parts of the scalar components of @xmath5 and thus possible inflaton directions .
the @xmath38-dependent mass terms generating a slope for the inflaton via the loop potential can be calculated from the scalar f - term potential and the fermionic mass matrix as before by @xmath84 \ ; \big|_{\hat{\phi } \rightarrow \phi } \ , , \\ ( { \cal m}_f)_{ij } & = e^{k/2}(w_{ij } + k_{ij}w + k_iw_j + k_j w_i + k_i k_j w - k^{k\overline{l}}k_{ij\overline{l } } { \cal d}_k w ) \ ; \big|_{\hat{\phi } \rightarrow \phi } \ , .
\end{split}\ ] ] the relevant contributions to the loop potential are given by the @xmath64 and @xmath10 mass terms which now obtain sugra corrections : @xmath85 \,,\\ ( m_h^{(p)})^2 & = 2 \,\kappa^2 m^2\left [ x + 1 + \left(\frac{m}{m_p}\right)^2 \left(1 - \kappa_{sh}\right)/2\right ] \,,\\ ( m_h^{(f)})^2 & = 2 \,\kappa^2 m^2 x \,,\\
( m_{h _ { a}}^{(s)})^2 & = ( m_{h _ { a}}^{(p)})^2 = \kappa^2\frac{m^4}{m_p^2 } + \frac{1}{2}\ , n^2 \sum_j |(y_{\nu})_{1j}|^2\,,\\ ( m_{h _ { a}}^{(f)})^2 & = \frac{1}{2}\ , n^2 \sum_j |(y_{\nu})_{1j}|^2\ , , \\ ( m_{l^j_a}^{(s)})^2 & = ( m_{l^j_a}^{(p)})^2 = \kappa^2\frac { m^4}{m_p^2 } + \frac{1}{2}\ , n^2 |(y_{\nu})_{1j}|^2\ , , \\
( m_{l^j_a}^{(f)})^2 & = \frac{1}{2}\ , n^2 |(y_{\nu})_{1j}|^2 \ , .
\end{split}\ ] ] comparing these expressions to the mass terms calculated in section [ susy ] we note some important points . a second mass scale , the scale of the sugra mass splitting @xmath86 , has appeared .
however this scale is much smaller than the susy mass splitting scale @xmath87 and thus we shall keep the latter scale as the renormalization scale .
the additional mass splitting implies that the @xmath10 and @xmath67 contributions no longer cancel .
however since the mass splitting is small compared to the susy mass splitting appearing in the @xmath13 mass terms and since the remaining parts of the @xmath10 and @xmath67 mass terms are proportional to @xmath88 these additional contributions to the loop potential are negligible for @xmath89 .
we will see later that this easily holds in our model .
furthermore , a new parameter has appeared in the loop potential : * the * sugra correction parameter * @xmath90 controls the sugra corrections to the loop potential .
@xmath91 recovers the phenomenology of global susy . as in the globally supersymmetric case , predictions for observables describing the cmb spectrum
can now be obtained by solving eq . and evaluating eqs . at the time when the cmb fluctuations exited the horizon .
in the sugra scenario , this was done numerically for @xmath92 . fixing the phase transition scale @xmath29 by the experimental value for the amplitude of the cmb fluctuations @xmath46 , the behavior of the spectral index @xmath47 , its running @xmath48 and the tensor - to - scalar ratio @xmath49
is shown in fig .
[ fig_pred ] . interpreting the results visualized in fig .
[ fig_pred ] and enforcing the experimental bounds of eqs
. implies restrictions on the model parameters . *
the * phase transition scale @xmath29 * is fixed to @xmath93 gev with a slight deviation in the region of large sugra corrections .
this is consistent with the global susy calculation ( from eqs . and ) which gives @xmath94 for @xmath95 .
* the width of the band in fig .
[ fig_pred ] is given by the variation of the * vacuum energy parameter @xmath17*. a priori we would expect @xmath17 to be an @xmath96 parameter , thus we shall assume @xmath97 . in fig .
[ fig_pred ] , larger values of @xmath17 are associated with sugra corrections becoming relevant at smaller values of @xmath98 .
in particular the tensor - to - scalar ratio @xmath49 is quite sensitive to @xmath17 with @xmath99 leading to comparatively larger @xmath100 .
[ fig_pred ] also demonstrates the effect of the * sugra correction parameter * @xmath90 .
the respective quantities are marked in black for @xmath101 which corresponds to the globally supersymmetric limit and in blue ( red ) for @xmath102 which corresponds to turning on the sugra corrections in the @xmath13 mass terms with positive ( negative ) sign . in the considered sugra context
the value of @xmath90 is a priori undetermined .
thus we would in general not expect to find global susy restored , which would correspond to @xmath90 exactly equal to one .
* the second parameter controlling the effect of the sugra corrections is the * seesaw scale * @xmath103 .
[ fig_pred ] shows that these corrections are suppressed for small @xmath98 , i.e. the observables are independent of @xmath104 for small values of the seesaw scale and the model predicts ( for @xmath29 fixed by the experimental value of @xmath46 ) @xmath105 note that in this case eqs .
for very small values of the seesaw scale @xmath98 , the spectral index @xmath47 approaches 1 , which is not preferred by the latest wmap data . on the other hand , all solutions with
@xmath106 leave the experimentally preferred region for the spectral index at large values of @xmath98 . in combination
, we find the preferred regions @xmath107 for @xmath108 and @xmath109 for @xmath110 , respectively .
for @xmath111 and @xmath112 , respectively . ] motivated by the above results , we take the phase transition scale to be fixed at @xmath113 and concentrate on the parameter ranges @xmath114 for the running of the spectral index and the tensor - to - scalar ratio .
after the end of the inflationary epoch the homogeneous classical fields and their quantum fluctuations evolve according to their respective equations of motion .
the universe enters a matter dominated regime until the decay of heavy particles and the thermalization of the light particles result in the total energy density being dominated by radiation ( fig .
[ fig_numerics ] ) .
this out - of - equilibrium decay of heavy particles can furthermore produce the necessary lepton asymmetry . in the following we will study these processes for the classical fields ( reheating ) and briefly comment on possible effects originating from their fluctuations ( preheating ) . to this end
we will start with the equations of motion for classical scalar fields , justify a simplification to boltzmann equations and finally derive analytical expressions for the generated baryon asymmetry and the reheat temperature .
we finish by commenting on preheating via parametric resonance and tachyonic preheating in this context .
the equations of motion for the scalar fields can be obtained by adding a phenomenological decay term @xcite to eq . thus giving @xmath115 adding a boltzmann equation for the quickly thermalizing direction
is protected against large sugra corrections .
] ultra - relativistic particles and the friedmann equation , we arrive at a closed set of differential equations : @xmath116 with @xmath117 , @xmath118 and @xmath119 denoting the energy densities of the ultra - relativistic particles , the @xmath38-field and the @xmath15-field respectively with @xmath120 .
having solved eq .
- , the lepton number density @xmath121 can be calculated from the boltzmann equation @xmath122 with the cp - violation per ( s)neutrino decay @xmath123 for a hierarchical spectrum of right - handed neutrinos bounded by @xcite @xmath124 here @xmath125 denotes the vacuum expectation value of the up - type higgs .
the lepton asymmetry is typically normalized to the entropy density @xmath126 @xcite with the effective number of degrees of freedom @xmath127 for the mssm particles .
the asymmetry @xmath128 is transferred to the baryon sector via sphaleron processes @xmath129 where @xmath130 is a number @xmath96 depending on the field content of the model and the temperature @xmath131 when the sphalerons leave equilibrium . in the mssm @xmath132
the quantity measured today is @xmath133 which can be calculated from the results above using the current conversion factor @xmath134 @xcite .
the second important physical quantity in the theory of reheating is the temperature of the universe when the universe becomes radiation dominated ( @xmath135 ) , the so - called reheat temperature .
it can be calculated from the results above using @xmath136 .
- assume that both the @xmath38-particles and the @xmath15-particles decay into ultra - relativistic particles with the respective decay rates @xmath137 and @xmath138 .
we will now describe a possibility how to evaluate these quantities in our framework . at the beginning of the reheating phase @xmath139 holds which implies that the decaying particles are damped predominantly by hubble expansion , not by decays , and the produced ultra - relativistic particles are strongly diluted .
the decays become significant for @xmath140 . at this stage
it is safe to assume @xmath141 . in this limit the respective decay rates derived from eq . are @xmath142 with the sneutrino @xmath38 decaying directly into lepton and higgsino or slepton and higgs and the @xmath15 particles decaying predominantly into the heaviest fermionic neutrino ( assuming this is not strongly suppressed by kinematics ) , .
the expressions in brackets emphasize that the decay is kinematically possible if @xmath143 .
however in the parameter range of interest , @xmath144 holds , as can be seen from eqs . .
thus we have @xmath145 . ]
which then in turn decays into lepton and higgs or slepton and higgsino .
note that the boltzmann equations and imply a splitting of the total matter energy density @xmath146 into @xmath118 and @xmath119 , which is not straightforward if the respective degrees of freedom are highly coupled .
however , since @xmath147 in our setting ( see section [ framework ] ) implies @xmath148 .
the assumption of hierarchical heavy neutrinos implies @xmath149 . ] in the preferred region of parameter space ( see section [ infl_lepto ] ) , any radiation energy density produced by @xmath15-decays will be strongly diluted during the following matter dominated phase governed by oscillations of the sneutrino . with @xmath150 shortly after the end of inflation due to the strong damping of the @xmath15-field ( see below ) we can thus substitute and by @xmath151 without introducing a significant error for the finally produced radiation density .
the effect of this approximation on the hubble expansion rate is negligible since the friedmann equation is predominantly governed by @xmath146 for @xmath152 . solving eqs .
, , , and numerically , we obtain the time evolution of scalar fields , the energy densities , the scale factor and the lepton asymmetry , respectively .
the former two are displayed in fig .
[ fig_numerics ] .
the regime of reheating is characterized by oscillating scalar fields and can be divided into distinct phases : after the end of inflation both @xmath38 and @xmath15 fall to their true minimum and begin to oscillate .
after only a few oscillations the classical field @xmath15 settles at its minimum and the dynamics of the system is governed by the oscillation of the @xmath38 field .
the further evolution of the @xmath38 oscillations is governed by hubble damping . as long as @xmath153 the universe
is governed by ( damped ) oscillating scalar fields which can be interpreted as ( decaying ) heavy particles .
this implies a matter dominated universe out of thermal equilibrium .
ultra - relativistic particles are produced through the decays of the heavy particles , however they are diluted by the expansion of the universe .
as soon as @xmath154 the radiation energy density becomes dominant and the light particles begin to thermalize .
this marks the end of reheating and determines the reheat temperature and the asymmetry @xmath155 .
since the set of equations , , and is quite involved , a common attempt in the literature ( e.g. @xcite ) is to simplify these equations by time - averaging the equations of motion of the scalar fields .
the result is a set of boltzmann equations for the matter energy density @xmath156 and the radiation energy density @xmath117 completed by the friedmann equation @xmath157 the lepton asymmetry is determined by eq . .
the big advantage is that these equations have approximate analytical solutions .
however their derivation ( see e.g. @xcite ) implies an important assumption concerning the scalar potential @xmath158 . in order to rewrite the time - averaged kinetic energy density in terms of the total energy density by exploiting the virial theorem
we must assume that we can write the scalar potential as @xmath159 with @xmath160 and @xmath161 .
- are obtained with @xmath162 .
numerical simulations of the full system , , , and show that this assumption is not justified in the early oscillation phase in the model described by eq . since the large oscillations of the @xmath38-field result in a highly coupled system with higher orders terms in the scalar potential playing a non - negligible role .
however they do also show that for @xmath163 the simpler system of differential equations - does give a good approximation .
this is the point of time relevant for the predictions of the reheating phase .
having seen that the results of the numerical solutions to the full field equations for @xmath164 can be approximated reasonably well by the simpler set of boltzmann differential equations , we can now find approximate analytical solutions to the latter and use these expressions to find estimates for the reheat temperature and the produced baryon asymmetry @xmath165 combining and reproduces the familiar relation @xmath166 ( see e.g. @xcite ) .
these results must be compared with existing bounds on the reheating process .
the wmap 7 year data combined with measurements of the baryon acoustic oscillations and todays hubble parameter imply @xmath167 @xcite , thus yielding @xmath168 gev . additionally , the reheat temperature is bounded from above by the so - called gravitino problem @xcite . a high reheat temperature would result in an overproduction of gravitinos .
if these are stable , then the fact that their energy density can not be larger than the present total energy density of the universe leads to a bound on the reheat temperature in terms of the gravitino mass @xmath169 . on the other hand , if gravitinos are not stable , they can either decay before or during and after the big bang nucleosynthesis ( bbn ) .
in the former case ( i.e. heavy gravitinos ) , with r - parity conserved the gravitinos will decay into the lightest supersymmetric particle ( lsp ) and their production is thus constrained by the dark matter abundance .
this yields a fairly model independent bound of @xmath170 gev for an lsp mass of about 100 gev to 150 gev . in the latter case
( i.e. light gravitinos ) , the decay of the gravitinos would alter the outcome of bbn and create a conflict between bbn predictions and observations .
this yields even stronger , however model dependent , constraints on the reheat temperature .
combining these arguments yields a constraint on the reheat temperature of typically @xmath171 , depending mainly on the model under consideration and on the value of @xmath169 .
the resulting preferred region in ( @xmath172)-parameter space is depicted in blue in fig .
[ fig_infl_lepto ] . note that throughout this chapter we have focussed on the evolution of the homogeneous fields @xmath38 and @xmath15 .
it has been pointed out that under certain circumstances this might not be sufficient , since modes with @xmath173 of all fields in the model can be strongly excited at the end of inflation and before the beginning of reheating in a process referred to as preheating .
there are two types of preheating worth mentioning in the context of hybrid inflation , namely preheating via parametric resonance @xcite and tachyonic preheating @xcite . in the former case , the coupling of fermions and bosons to the oscillating inflaton field results in oscillating mass terms for these particles .
solving the respective equations of motion ( roughly the equation of an harmonic oscillator with an oscillating mass as described , e.g. , by the mathieu equation ) can yield explosive particle production .
however , in the region of parameter space of interest to us , any heavy particles that are produced by this mechanism will decay back into heavy ( s)neutrinos or into radiation . the radiation produced directly or indirectly through this process at the beginning of the reheating phase
will however be strongly diluted during the ongoing matter dominated phase and thus be insignificant for the outcome of the reheating phase .
thus in our model , parametric resonance will not affect the results discussed above , mainly due to the structure of the mass spectrum and the very small effective yukawa coupling @xmath174 .
tachyonic preheating occurs when the squared mass of the @xmath15 field becomes negative , triggering the waterfall ending inflation .
modes of the @xmath15 field with @xmath175 grow exponentially , potential ( with @xmath176 ) , which does not appear in our model as can easily be checked from eq . .
] causing particle production of bosonic and fermionic fields coupled to the waterfall field @xcite and creating an inhomogeneous field @xmath177 which can cause the formation of topological defects when the waterfall occurs @xcite .
it was stated in @xcite that the production of fermions and bosons coupling to the waterfall field with a coupling strength @xmath178 is suppressed by @xmath179 with @xmath180 denoting the energy density during inflation .
thus in the parameter region of interest , this is negligible in our model . on the other hand
, the production of topological defects could indeed dominate the evolution of the universe in an early stage .
however , since we have not observed any topological defects yet , a mechanism to prevent or dilute these objects ( e.g. a preferred waterfall direction or a slight shift of the potential energy of the discrete vacua ) is typically implemented .
we will assume that the higher dimensional operators denoted by dots in eq .
provide such a solution so that at some time after the waterfall , the universe is dominated by the lightest right - handed sneutrino . the evolution from this point on
is correctly described by the classical theory of reheating , as discussed above .
other possible scenarios in which the evolution of the universe may not be dominated by the homogeneous component of the inflaton field remain to be explored in this context .
in sections [ inflation ] and [ leptogenesis ] we have investigated the conditions under which inflation , with primordial perturbations in accordance with the latest wmap results , as well as successful leptogenesis can be realized simultaneously in simple models of sneutrino hybrid inflation as outlined in section [ framework ] .
the combined results are summarized in fig .
[ fig_infl_lepto ] .
the dynamics of inflation is governed by the scale @xmath29 of the phase transition ending hybrid inflation , the mass of the lightest right - handed ( s)neutrino @xmath98 , the vacuum energy parameter @xmath17 (= waterfall field self coupling ) and the parameter @xmath90 controlling the sugra corrections . in principle , terms depending on the neutrino yukawa coupling matrix could contribute , too . however
, in our case the comparatively small first generation yukawa couplings make these contributions negligible . with @xmath29 fixed by the amplitude of the scalar cmb fluctuations and @xmath181 , the spectrum of the cmb fluctuations is primarily dependent on the lightest right - handed ( s)neutrino mass @xmath98 . for large values of @xmath98 , sugra corrections controlled by
@xmath90 become important , with the sign of these contributions depending on the sign of @xmath182 . for the spectral index , its running and the tensor - to - scalar ratio the predictions
are shown in fig .
[ fig_pred ] .
recent wmap observations constrain the preferred region for the spectral index @xmath47 , thus imposing a constraint on the preferred region for @xmath98 . for @xmath183 and @xmath184
this is marked in grey in fig .
[ fig_infl_lepto ] .
on the other hand , the decisive quantities of reheating and leptogenesis , namely the reheat temperature @xmath185 and the baryon asymmetry @xmath186 depend on the effective first generation neutrino yukawa coupling @xmath174 , the cp asymmetry @xmath187 and the mass of the lightest right - handed ( s)neutrino @xmath98 ( see eqs . and ) .
the latter parameter is thus the link between inflation and leptogenesis .
the preferred region of parameter space resulting from bounds on these quantities is marked in blue in fig .
[ fig_infl_lepto ] .
it is bounded from below by the experimental value of the baryon asymmetry measured by wmap and by an upper bound on the cp - violation per ( s)neutrino decay eq . .
from above it is bounded by constraints imposed on the reheat temperature from the gravitino problem .
since these are model dependent , we have plotted the regions satisfying @xmath188 gev in different shadings .
note that a higher reheat temperature at a fixed value for @xmath98 automatically corresponds to a smaller value of @xmath187 in order to match the measured baryon asymmetry .
the resulting preferred region in parameter space implies an effective first generation yukawa coupling @xmath189 .
the upper part of this range is of the same order as the first family quark and charged lepton yukawa couplings , which in the mssm with moderate @xmath190 are of the order @xmath191 to @xmath192 . throughout this paper
, we have assumed nonthermal leptogenesis and hierarchical masses of left - handed as well as right - handed neutrinos .
assuming that the light neutrinos obtain masses via a type i seesaw mechanism . ] , both assumptions depend on the value of the effective light neutrino mass parameter ( also dubbed washout parameter ) @xmath193 .
more explicitly , one can easily see from eq . that @xmath194 .
lines of constant @xmath195 are marked in red in fig .
[ fig_infl_lepto ] , corresponding to a fixed relation between @xmath185 and @xmath98 .
simultaneously , they give the order of magnitude for the mass of the left - handed neutrino @xmath196 . in the preferred region of parameter space
, we find @xmath197 ev , thus implying nonthermal leptogenesis with @xmath198 and @xmath199 .
finally , we want to comment on possible extensions of this scenario and the significance of cosmological observations in the near future . in fig .
[ fig_infl_lepto ] we have set @xmath183 . allowing for @xmath97 gives qualitatively the same picture ( see section [ inflation ] , in particular fig
. [ fig_pred ] ) with a somewhat shifted grey region .
for example , for @xmath99 the grey region is extended to the left to @xmath200 gev whereas for @xmath201 it is extended to the right to @xmath202 gev for @xmath203 .
the respective region is extended to @xmath204 gev . ]
another interesting possibility would arise if the experimentally preferred region for the spectral index was raised , favoring a spectral index closer to 1 .
this would lower the preferred range for the lightest ( s)neutrino mass @xmath98 significantly and thus open up the region of thermal leptogenesis and allow for @xmath205 .
the forthcoming results of the planck satellite will make the requirements for @xmath98 more accurate . in summary ,
we have pointed out that successful sneutrino hybrid inflation and leptogenesis can be achieved in this framework , and that combining both imposes requirements on the parameters of the underlying particle physics model .
we obtain a mass for the lightest right - handed ( s)neutrino of @xmath206 gev , an effective first generation neutrino yukawa coupling @xmath189 and a very light left - handed neutrino with @xmath207 ev .
furthermore , we find that leptogenesis occurs via nonthermal leptogenesis ( with @xmath208 for @xmath183 ) . concerning the spectrum of the cmb fluctuations , we predict a running of the spectral index of @xmath209 and a tensor - to - scalar ratio of @xmath210 .
our results provide a guideline for the construction of explicit particle physics models incorporating sneutrino hybrid inflation and subsequent baryogenesis via nonthermal leptogenesis .
we would like to thank alejandro ibarra and koushik dutta for discussions .
we acknowledge partial support by the dfg cluster of excellence `` origin and structure of the universe '' .
a. h. guth , phys .
* d23 * ( 1981 ) , 347356 ; a. d. linde , phys . lett .
* b108 * ( 1982 ) , 389393 ; a. albrecht and p. j. steinhardt , phys .
* 48 * ( 1982 ) , 12201223 ; for a review containing an extensive list of references , see e.g. : d. h. lyth and a. riotto , phys .
rept . * 314 * ( 1999 ) 1 . for textbook reviews on inflation
see : a. r. liddle and d. h. lyth , `` cosmological inflation and large - scale structure , '' _ cambridge , uk : univ .
( 2000 ) 400 p _
; a. d. linde , `` particle physics and inflationary cosmology , '' [ arxiv : hep - th/0503203 ] ; v. mukhanov , `` physical foundations of cosmology , '' _ cambridge , uk : univ . pr .
( 2005 ) 421 p _ m. fukugita and t. yanagida , phys .
b * 174 * , 45 ( 1986 ) .
p. minkowski , phys .
b * 67 * ( 1977 ) 421 ; m. gell - mann , p. ramond and r. slansky in sanibel talk , calt-68 - 709 , feb 1979 , and in _ supergravity _ ( north holland , amsterdam 1979 ) ; t. yanagida in _ proc . of the workshop on unified theory and baryon number of the universe _ , kek , japan , 1979 ; s.l.glashow , cargese lectures ( 1979 ) ; r. n. mohapatra and g. senjanovic , phys .
* 44 * ( 1980 ) 912 ; j. schechter and j. w. valle , phys .
d * 25 * ( 1982 ) 774 .
m. c. chen , arxiv : hep - ph/0703087 .
s. davidson , e. nardi and y. nir , phys .
rept . * 466 * , 105 ( 2008 ) [ arxiv:0802.2962 [ hep - ph ] ] . h. murayama , h. suzuki , t. yanagida and j. yokoyama , phys .
* 70 * ( 1993 ) 1912 .
j. r. ellis , m. raidal and t. yanagida , phys .
b * 581 * , 9 ( 2004 ) [ arxiv : hep - ph/0303242 ] .
s. antusch , m. bastero - gil , s. f. king and q. shafi , phys .
d * 71 * , 083519 ( 2005 ) [ arxiv : hep - ph/0411298 ] .
a. d. linde , phys .
b * 249 * , 18 ( 1990 ) .
a. d. linde , phys .
b * 259 * , 38 ( 1991 ) .
e. j. copeland , a. r. liddle , d. h. lyth , e. d. stewart and d. wands , phys .
d * 49 * , 6410 ( 1994 ) [ arxiv : astro - ph/9401011 ] .
a. d. linde and a. riotto , phys .
d * 56 * , 1841 ( 1997 ) [ arxiv : hep - ph/9703209 ] .
s. antusch , k. dutta and p. m. kostka , aip conf .
* 1200 * , 1007 ( 2010 ) [ arxiv:0908.1694 [ hep - ph ] ] .
m. kawasaki , m. yamaguchi and t. yanagida , phys .
lett . * 85 * , 3572 ( 2000 ) [ arxiv : hep - ph/0004243 ] . m. yamaguchi and j. yokoyama , phys .
d * 63 * , 043506 ( 2001 ) [ arxiv : hep - ph/0007021 ] .
m. kawasaki , m. yamaguchi and t. yanagida , phys .
d * 63 * , 103514 ( 2001 ) [ arxiv : hep - ph/0011104 ] .
s. antusch , k. dutta and p. m. kostka , phys .
b * 677 * , 221 ( 2009 ) [ arxiv:0902.2934 [ hep - ph ] ] . s. antusch , m. bastero - gil , k. dutta , s. f. king and p. m. kostka , jcap * 0901 * , 040 ( 2009 ) [ arxiv:0808.2425 [ hep - ph ] ] . s. antusch , m. bastero - gil , j. p. baumann , k. dutta , s. f. king and p. m. kostka , arxiv:1003.3233 [ hep - ph ] . s. antusch , s. f. king , m. malinsky , l. velasco - sevilla and i. zavala , phys .
b * 666 * , 176 ( 2008 ) [ arxiv:0805.0325 [ hep - ph ] ] .
w. j. percival
_ et al . _ , mon . not .
* 401 * , 2148 ( 2010 ) [ arxiv:0907.1660 [ astro-ph.co ] ] .
a. g. riess _ et al .
_ , astrophys .
j. * 699 * , 539 ( 2009 ) [ arxiv:0905.0695 [ astro-ph.co ] ] .
wmap cosmological parameters .
model : lcdm+sz+lens .
data : wmap7+bao+h0 + http://lambda.gsfc.nasa.gov/product/map/current/params/ lcdm@xmath211sz@xmath211lens@xmath211wmap7@[email protected] l. kofman , a. d. linde and a. a. starobinsky , phys .
d * 56 * , 3258 ( 1997 ) [ arxiv : hep - ph/9704452 ] .
r. allahverdi and a. mazumdar , jcap * 0610 * , 008 ( 2006 ) [ arxiv : hep - ph/0512227 ] .
r. allahverdi and a. mazumdar , phys .
d * 76 * , 103526 ( 2007 ) [ arxiv : hep - ph/0603244 ] .
l. covi , e. roulet and f. vissani , phys .
b * 384 * ( 1996 ) 169 [ arxiv : hep - ph/9605319 ] .
s. davidson and a. ibarra , phys .
b * 535 * , 25 ( 2002 ) [ arxiv : hep - ph/0202239 ] .
k. hamaguchi , h. murayama and t. yanagida , phys .
d * 65 * , 043512 ( 2002 ) [ arxiv : hep - ph/0109030 ] .
e. w. kolb and m. s. turner , front .
phys . * 69 * , 1 ( 1990 ) .
e. komatsu _ et al .
_ , arxiv:1001.4538 [ astro-ph.co ] . m. y. khlopov and a. d. linde , phys . lett .
b * 138 * , 265 ( 1984 ) .
j. r. ellis , j. e. kim and d. v. nanopoulos , phys .
b * 145 * , 181 ( 1984 ) .
j. r. ellis , d. v. nanopoulos and s. sarkar , nucl .
b * 259 * , 175 ( 1985 ) .
t. moroi , h. murayama and m. yamaguchi , phys .
b * 303 * , 289 ( 1993 ) .
m. kawasaki , k. kohri and t. moroi , phys .
b * 625 * , 7 ( 2005 ) [ arxiv : astro - ph/0402490 ] .
m. bastero - gil , s. f. king and j. sanderson , phys .
d * 60 * , 103517 ( 1999 ) [ arxiv : hep - ph/9904315 ] .
j. garcia - bellido , s. mollerach and e. roulet , jhep * 0002 * , 034 ( 2000 ) [ arxiv : hep - ph/0002076 ] .
j. garcia - bellido and e. ruiz morales , phys .
b * 536 * , 193 ( 2002 ) [ arxiv : hep - ph/0109230 ] .
g. n. felder , j. garcia - bellido , p. b. greene , l. kofman , a. d. linde and i. tkachev , phys .
lett . * 87 * , 011601 ( 2001 ) [ arxiv : hep - ph/0012142 ] .
j. mcdonald , phys .
d * 66 * , 043525 ( 2002 ) [ arxiv : hep - ph/0105235 ] . | in sneutrino hybrid inflation the superpartner of one of the right - handed neutrinos involved in the seesaw mechanism plays the role of the inflaton field .
it obtains its large mass after the `` waterfall '' phase transition which ends hybrid inflation . after this phase transition
the oscillations of the sneutrino inflaton field may dominate the universe and efficiently produce the baryon asymmetry of the universe via nonthermal leptogenesis .
we investigate the conditions under which inflation , with primordial perturbations in accordance with the latest wmap results , as well as successful nonthermal leptogenesis can be realized simultaneously within the sneutrino hybrid inflation scenario .
we point out which requirements successful inflation and leptogenesis impose on the seesaw parameters , i.e. on the yukawa couplings and the mass of the right - handed ( s)neutrino , and derive the predictions for the cmb observables in terms of the right - handed ( s)neutrino mass and the other relevant model parameters .
0.05 cm * sneutrino hybrid inflation and + nonthermal leptogenesis * stefan antusch , jochen p. baumann , valerie f. domcke and + philipp m. kostka + _ max - planck - institut fr physik ( werner - heisenberg - institut ) , + fhringer ring 6 , 80805 mnchen , germany _ 0.75 cm |
in double quantum wells , excitons are more complex than in a single quantum well , because the electron and hole wave - functions can be localized either in the same well , or in different wells , or in both .
the asymmetry of the quantum well produces transitions which are prohibited in the optical spectrum , of the symmetric case .
the theoretical study of excitons in double quantum wells , began with the work of kamizato and matsuura @xcite .
they studied the excitonic levels and binding energy of symmetric wells as a function of the wells and barrier widths .
dignam and sipe @xcite improved the calculation for the excitonic levels behavior for wide barriers , incorporating the coulomb interaction between excitonic states .
their work included both symmetric and asymmetric wells , and the presence of an external electric field ( see also ref . ) , but their method of calculation is not suitable for thin barriers .
cen and bajaj @xcite improved the variational method used in ref .
, obtaining the exciton binding energy , in symmetric wells in an external magnetic field , for arbitrary wells and barrier width . recently dzyubenko and yablonskii @xcite , using a non variational method , studied the excitonic levels for symmetric and almost symmetric double quantum wells , as a function of the electric and magnetic field strength . for an experimental study of excitonic levels in @xmath2 symmetric double quantum wells
see bayer @xmath3 @xcite , and for a recent review in the subject of excitons in double quantum wells , see ref . .
in this work , we studied the excitonic levels , binding energy and optical transition probabilities , in symmetric and asymmetric double quantum wells , in the presence of a magnetic field , as a function of barrier and well widths and the magnetic field strength .
the interaction between light and heavy holes @xcite is not considered in our work and we restrict the analysis only to excitons composed by electrons and heavy holes . within the effective mass approach ,
we expand the excitonic wave - function , in a orthogonal basis formed by products of electron and hole wave - functions in the growth direction @xmath0 , and one particle solutions of the magnetic hamiltonian in the @xmath1 plane .
the coulomb potential between electrons and holes produces off - diagonal terms by mixing our basis states .
we obtained the energies and wave - functions by diagonalizing the excitonic hamiltonian in a truncated basis .
in contrast to the majority of works in the literature @xcite , our basis set is orthogonal @xcite , and we do not use a variational method .
we applied our method to study the first excitonic states , in @xmath4 ( @xmath2 ) heterostructures . for comparison purposes in the symmetric case
, we used the parameters of ref . .
our analysis of the basis states involved in the excitonic wave - function for different quantum wells geometries , should be helpful in similar studies of @xmath5 and @xmath6 @xcite systems .
the effective mass hamiltonian for excitons in a double quantum well in the presence of a magnetic field pointing towards @xmath0 , in the diagonal approximation @xcite can be written as @xmath7 @xmath8 is the 1-dimensional hamiltonian for electrons , @xmath9 @xmath10 is the 1-dimensional hamiltonian for holes , @xmath11 and @xmath12 ( @xmath13 ) is the potential that defines the double quantum well for electrons ( holes ) in the five regions of @xmath0 , shown in fig .
[ pot - z ] .
@xmath14 is the magnetic hamiltonian in the symmetric gauge , which depends on the relative coordinates of electrons and holes in the @xmath1 plane , @xmath15 where @xmath16 , @xmath17 and @xmath18 are the momentum operator , reduced mass and hole mass , defined in the @xmath1 plane .
@xmath19 is the coulomb potential between electrons and holes , including an effective dielectric constant for the system .
we expanded the solution of the hamiltonian [ h - total ] , as a lineal combination of products of eigenfunctions of the magnetic hamiltonian in the @xmath1 plane ( [ h - mag ] ) , and eigenfunctions of the electron and hole hamiltonians in the @xmath0 direction ( [ h - ele ] and [ h - hole ] ) , @xmath20 in which @xmath21 where only @xmath22 functions are considered , and @xmath23 . the electron wave - function defined in the five regions of @xmath0 , shown in fig .
[ pot - z ] , is given by @xmath24 and the hole wave - function is given by similar expressions .
the coulomb interaction produces off - diagonal terms by mixing our basis states . in order to obtain the system of equations for the coefficients in the expansion ( [ expansion ] ) , we need to evaluate the coulomb integral @xmath25 the @xmath26 integral is trivial , because of @xmath27 conservation . using the explicit expansion of laguerre polynomials ( @xmath28 ) in @xmath29 and @xmath30 ( [ f - mag ] ) , the remaining of integral ( [ i - total ] )
can be written as a sum of terms of the form : @xmath31 by using @xmath32 and after solving the @xmath33 integral , it yields @xmath34 the @xmath35 and @xmath36 integrals can be solved analytically .
the evaluation of these integrals is cumbersome due to the large number of terms , resulting from the five different regions of the potential .
the @xmath35 and @xmath36 integrals contain both , decoupled terms in which the @xmath35 and @xmath36 integrals are independent of each other , and coupled terms where the integration limits of the @xmath36 integral contains @xmath35 .
the remaining @xmath37 integral must be calculated numerically .
diagonalizing the system of equations resulting for the coefficients in the expansion ( [ expansion ] ) , in a truncated basis , we obtained the energies and wave - functions for the first excitonic levels .
evaluating the oscillator strength @xmath38 we obtained the optical absorption spectra .
we calculated the excitonic energy levels and the magneto - optical absorption spectra for symmetric and asymmetric double quantum wells , composed by @xmath4 wells and @xmath2 barriers ( @xmath39 ) .
the band gap used in our calculations is given by @xmath40 .
the band gap offset considered was a sixty percent for the conduction band and a forty percent for the valence band . for comparison purposes in the symmetric case
, we used the same simplified masses of ref .
, for all five regions in the double quantum well .
the electron mass considered was @xmath41 , the @xmath1 plane heavy hole mass , @xmath42 , the @xmath43 heavy hole mass @xmath44 and the light hole masses @xmath45 , @xmath46 , which corresponds to @xmath47 and @xmath48 .
we considered a dielectric constant @xmath49 .
it was enough to consider in our calculations a truncated basis composed of twelve landau wave - functions , four electronic wave - functions and four heavy hole wave - functions .
in this section we present our results for the excitonic energy levels in symmetric double quantum wells as a function of wells and barrier widths , ranging from zero to large values , in a @xmath50 magnetic field .
also , we present the optical absorption spectra , including broadening for each possible optical transition .
[ binding]a shows the binding energy of the ground state , for a @xmath51 barrier and for both wells widths ranging from @xmath52 to @xmath53 . for zero well width , the exciton is tridimensional and it corresponds to bulk @xmath54 . for non - zero
well width , the exciton is localized within the wells , reaching a maximum binding energy at a certain well width which corresponds to a maximum two - dimensional character . for further increase in the wells width , the binding energy decreases and the exciton evolves towards a tridimensional @xmath4 exciton for thick wells .
this behavior of the ground state compares very well with the results of ref . , although our energies are @xmath55 lower .
[ binding]b shows the ground state binding energy for wells width fixed in @xmath56 , and barrier width ranging from @xmath52 to @xmath56 . as the barrier width
is varied from zero to infinity , we move from the case of a single quantum well ( width : @xmath57 ) , towards two independent quantum wells of width @xmath58 and @xmath59 . it is known that when the system size decreases , the exciton binding energy increases , accordingly the binding energy starts in a value which corresponds to a single quantum well of width @xmath57 and ends in a larger value that corresponds to each decoupled quantum well .
the binding energy decreases for thinner barriers , because the wave - functions have a large amplitude within the barrier , which is equivalent to a single quantum well wider than @xmath57 . in this case
we obtain an excellent agreement with ref . .
[ sym - lb]a shows the ground state and higher excitonic levels in a symmetric double quantum well , for @xmath56 wells and barrier width ranging from @xmath52 to @xmath60 .
it can be seen that the second energy level ( optically prohibited in the symmetric case ) , gets closer to the ground state for wider barriers .
this is a typical behavior , when we go from a situation of coupled quantum wells towards one of two decoupled wells having degenerate energies . in this case
the coulomb potential do not mix states with different symmetry under simultaneous interchange of @xmath35 by @xmath61 and @xmath36 by @xmath62 . in terms of our base wave - functions ,
the first two excitonic states evolve from @xmath63 ( first electron - first hole , exciton ) and @xmath64 ( first electron - second hole , exciton ) , towards @xmath65 and @xmath66 as the barrier get wider .
this behavior was noted ( without magnetic field ) in ref .
and corresponds to the ground state evolving towards @xmath67 , in terms of a basis of single quantum well wave - functions , corresponding to the electron ( or hole ) localized in the left well @xmath68 or the right well @xmath69 .
this classification in terms of @xmath68 and @xmath69 states , will be useful in explaining the optical absorption spectra for thick barriers . in fig .
[ sym - lb]b , we show the optical absorption spectra for @xmath56 wells , with the central barrier ranging from @xmath52 to @xmath70 .
the state @xmath71 evolves toward @xmath72 getting close to the state @xmath65 .
this state disappear from the spectra , because is evolving towards @xmath73 , where the coupling between electron and hole wave - functions decreases , as the electron and the hole are localized in separated wells .
the finite energy separation between these states for wide barriers , is a consequence of the small binding energy for the @xmath72 state , and corresponds to the binding energy for the @xmath65 state for very large barriers . for thin barriers , one small peak between @xmath63 and @xmath71 , corresponding to @xmath74 ,
can not be distinguished .
the other peaks correspond to higher levels and the shoulder in @xmath71 , which evolves growing in size , corresponds to a landau level of @xmath63 .
the states involved in the anti - crossing ( produced for a @xmath75 barrier ) have a very small oscillator strength , compared to the @xmath71 state . because of this , the anti - crossing manifest itself only as small distortion , in the evolution from @xmath71 towards @xmath72 , which finally disappear from the spectra for thick barriers , as mentioned previously . fig .
[ sym - lw]a shows the ground and higher excitonic states ( with respect to the first band to band transition ) for a symmetric double quantum well , with a central barrier of @xmath51 and both wells ranging from @xmath52 to @xmath53 .
the higher levels show an anticrossing region involving the states @xmath76 ( @xmath71 ) and a landau level .
the states @xmath63 and @xmath64 evolve towards @xmath65 y @xmath66 .
[ sym - lw]b shows the calculated optical absorption for the previous case .
the anticrossing region ( for @xmath77 wells ) manifests as a reinforcement in the spectral lines , caused by the coincidence of the landau energy and the energy separation between the @xmath63 and @xmath71 states .
it can be seen as the @xmath71 state get close to the @xmath63 state and loose probability for wide wells .
previous works only included the coupling between levels , necessary for having the right energy behavior for the excitonic wave - function for thick barriers , and the general belief was that coupling between others levels is of no importance . in our calculation is clear that although this is true for almost all wells and barrier widths , there are regions where another off - diagonal terms must be included , for explaining the anti - crossings that we obtained between higher levels ( see fig .
[ sym - lb]a ) . when going from the symmetric case towards the asymmetric one , states prohibited by the quantum well symmetry can appear in the optical absorption spectra .
these states have a small oscillator strength , near the symmetric case , which increases when the @xmath0 potential is going far from symmetry . in asymmetric double quantum wells , the excitonic ground state wave - function is mostly localized in the wider well . for @xmath58 smaller than @xmath59
the ground state binding energy corresponds to an exciton localized in the right well .
as @xmath58 get closer to @xmath59 , the exciton begin to be present in the left well and the binding energy decreases .
when the two wells are of the same width , the exciton is localized in both wells ( symmetric case ) and the binding energy reach a local minima . for @xmath58
bigger than @xmath59 , the exciton exciton is localized in the left well and the binding energy increases .
after this , the binding energy reach a maxima and decreases towards the tridimensional limit of a very thick left well .
[ asym - lw1]a shows the ground and higher excitonic states , for a @xmath56 right well , a @xmath78 barrier and a @xmath50 magnetic field , for the left well ranging from @xmath79 to @xmath56 .
the ground state remains fixed in energy and corresponds to an exciton localized in the right well ( the ground state binding energy behavior , explained previously , can not be observed in the scale of fig .
[ asym - lw1 ] ) .
the second excitonic state , prohibited and degenerate in energy with the ground state ( for thick barriers ) in the symmetric case , separates from the ground state when the left well width decreases . for a @xmath80 left well
, this state interacts with the third state , producing an anti - crossing . in the anticrossing region , these states lose their identity , transforming into @xmath81 and @xmath82 . from analyzing the energies and wave - functions in our basis states
, we see that this anti - crossing is not originated by a crossing between electron ( or hole ) levels of the right and left wells ( see ref . ) , as for almost all left well widths of fig .
[ asym - lw1]a , the electron ( hole ) levels @xmath83 ( @xmath84 ) are already localized in the right well and @xmath85 ( @xmath86 ) in the left well .
these anti - crossings , are a consequence of the strong interaction between excitonic states , when the levels corresponding to the direct exciton @xmath71 ( @xmath87 ) and indirect exciton @xmath64 ( @xmath88 ) , get close in energy . the different behavior in these energy levels , when one of the wells width is varied , is originated by the different spatial character of excitons localized in only one well ( direct ) and excitons where electron and hole are localized in different wells ( indirect ) . as can be seen from fig .
[ asym - lw1]b , both excitonic states have a large oscillator strength in the anti - crossing region , and because the energy separation is approximately @xmath89 between them , this anti - crossing could be observed experimentally . in fig .
[ asym - lb - b]a we show the excitonic states for a @xmath80 left well and a @xmath56 right well in a @xmath50 magnetic field , as a function of the barrier width .
it is clear the way the anti - crossing states evolve and get separated in energy as the barrier is decreased .
when these energy levels go far from each other , the probability of being created optically decreases .
this would make difficult to obtain experimentally this anti - crossing , for barriers smaller than @xmath78 . for wider barriers ,
the direct exciton @xmath71 has a lower energy than the indirect exciton @xmath64 , as a consequence of the large binding energy of @xmath71 , relative to @xmath64 .
these different binding energies are a consequence of the large spatial separation of the wave - functions corresponding to @xmath90 and @xmath91 . these anti - crossings has not be obtained previously .
we think that it is important to carry out an experimental study , of the correlation in the optical absorption spectrum , for the energies of the second and third excitonic states , in the anti - crossing region . also could be interesting , to study the possible relationship of the states @xmath81 and @xmath82 with charged excitons @xcite , because these states involve one particle wave - functions , corresponding to two electrons and one hole states .
increasing the magnetic field in a symmetric or asymmetric double quantum well , produces a shift in the excitonic energy levels towards higher energies and an increase in the binding energy .
the increase in the binding energy is a consequence of two effects : first , the wave - function confinement in the @xmath1 plane produces a stronger interaction in the @xmath43 , as the electron and hole wave - functions penetrates more in the barrier .
second , the binding between electron and hole in the plane increases , because their wave - functions are confined to a smaller region .
the large binding between electron and hole obtained when increasing the applied magnetic field , is similar to a change in the barrier and well width .
this can be used to study these systems in regions of interest , without the need for the growth of many different samples . in fig .
[ asym - lb - b]b , we show the magnetic field effects for the asymmetric double quantum well , in the anti - crossing region of fig .
[ asym - lb - b]a .
when the magnetic field ranges from @xmath52 to @xmath92 , the energy levels are shifted towards a bigger energy and the binding is stronger .
the behavior for the indirect excitonic levels is similar to decreasing the barrier width in the anti - crossing region .
the direct exciton binding energy also increases , as a consequence of the wave - function localization in the @xmath1 plane .
because the direct exciton has a larger binding energy than the indirect exciton for large magnetic fields , this level goes below the indirect exciton level for strong magnetic fields .
in this work we studied the energies and oscillator strength , for several excitonic @xmath93 levels in a double quantum well , in a magnetic field pointing in the growth direction @xmath0 .
we calculated the excitonic binding energy and optical absorption for type @xmath94 symmetric and asymmetric double quantum wells , as a function of barrier and well ( or wells ) widths . in our method
we used an orthogonal basis and we do not employed a variational method .
specific calculations for the ground state excitonic binding energy , in symmetric double quantum wells , are in excellent agreement with the work of cen - bajaj@xcite .
in contrast to the dignam - sipe@xcite work , we used basis functions of single particle solutions of the double quantum well , and our method is valid for all barrier width .
to our understanding , this is the first theoretical calculation , in a symmetric and asymmetric double quantum well , that is valid for all barrier widths .
this allow us to study these systems , from the case of a single quantum well ( null barrier ) to the case where both wells get decoupled .
we do not know another theoretical calculation that predict the anti - crossings between excitonic levels , that we obtain as a consequence of including in our method all the coupling between our basis wave - functions .
we want to acknowledge the hospitality of the people in the electronic department in glasgow university , specially to c. sotomayor , d. hutchings and j. arnold .
we also want to acknowledge a conicyt - british council grant for a research student - ship given to francisco vera , and to g. fuster for many discussions on this topic .
this work was supported by the fondecyt projects 1950190,1970119 and a conicyt doctoral fellowship .
t. kamizato and m. matsuura , phys .
b * 40*,8378 ( 1989 ) .
m. dignam and j. sipe , phys .
b * 43*,4084 ( 1991 ) .
r. ferreira , c .
delalande , h .
w. liu , g .
bastard , b. etienne and j. f. palmier , phys .
b * 42*,9170 ( 1990 ) . j.cen and k. bajaj , phys . rev .
b * 46*,15280 ( 1992 ) . a. dzyubenko and a. yablonskii , phys .
b * 53*,16355 ( 1996 ) .
m. bayer , v .
b. timofeev , f .
faller , t .
gutbrod and a. forchel , phys .
b * 54*,8799 ( 1996 ) .
r. ferreira and g. bastard , rep .
. phys . * 60*,345 ( 1997 ) .
a. hernandez and a. ramos , j. appl .
phys . * 80*,1547 ( 1996 ) .
i. linnerud and k. chao , phys .
b * 49*,8487 ( 1994 ) . s. ten , m .
f. krol , p .
t. guerreiro and n. peyghambarian , appl .
. lett . * 69*,3387 ( 1996 ) .
h. haas , n .
magnea and l. dang , phys .
b * 55*,1563 ( 1997 ) .
g. bastard , wave mechanics applied to semiconductor heterostructures , les editions de physique ( 1990 ) , p. 139 .
a. j. shields , j . l. osborne , d .
m. whittaker , m. y. simmons , m .
pepper and d. a. ritchie , phys .
b * 55*,1318 ( 1997 ) . g. finkelstein , h .
shtrikman and i. barjoseph , surface science * 362*,357 ( 1996 ) .
p. lelong and g. bastard , solid state commun .
* 98*,819 ( 1996 ) . | we studied theoretically the excitonic energy levels and the optical absorption spectra for double quantum wells , both symmetric and asymmetric , in the presence of an homogeneous magnetic field . within the effective mass approach
, we expanded the excitonic wave - function , in an orthogonal basis formed by products of electron and hole wave - functions in the growth direction @xmath0 , and one particle solutions of the magnetic hamiltonian in the @xmath1 plane .
we applied our method to the case of @xmath2 , for which we showed how the exciton wave - function vary , and how the basis functions are mixed in a non trivial way by the effect of the coulomb potential . by taking into account all the mixing between the elements in our base , we get anti - crossings between excited excitonic states not reported previously . |
let @xmath1 be an embedded minimal surface in @xmath2
. by the monotonicity formula , the area density @xmath3 is nondecreasing .
if @xmath4 we say that @xmath1 has quadratic area growth , or the area growth of @xmath5 planes . for surfaces with the growth of 2 planes , there are two canonical examples : the catenoid ( fig [ fig : catenoid ] ) , and scherk s singly periodic surfaces , which occur in a one parameter family ( fig [ fig : scherk ] and fig [ fig : scherk1 ] ) , where the parameter is the angle betwee the two leaves .
as the angle goes to zero , the scherk surfaces approach a catenoid on compact sets after an appropriate rescaling . in 2005
, meeks and wolf proved the following theorem : @xcite[thm : periodic ] suppose that @xmath1 is an embedded minimal surface in @xmath2 which has infinite symmetry group and @xmath6 .
then @xmath1 is either a catenoid or a scherk example .
meeks has conjectured that the symmetry condition in the above may be removed : @xcite let @xmath1 be an embedded minimal surface in @xmath2 with area growth of 2 planes . then @xmath1 is either a catenoid or a scherk example .
however , an initial difficulty with the above is that it is not yet known that a minimal surface with quadratic growth even needs to be asymptotic to a catenoid or a scherk example . by the compactness theory from gmt ,
it is known that if @xmath1 is an embedded minimal surface with quadratic area growth , then for any sequence @xmath7 , there exists a subsequence @xmath8 such that @xmath9 converges to a minimal cone @xmath10 in the varifold topology .
such a cone @xmath10 is called a * tangent cone at infinity*. a priori , there may be many tangent cones at infinity .
this leads to the following conjecture , also due to meeks : @xcite let @xmath1 be an embedded minimal surface in @xmath2 with quadratic area growth
. then @xmath1 has a unique tangent cone at infinity . in the case of finite genus , this had already been resolved by collin @xcite , who proved that any minimal surface with finite genus and quadratic area growth must be asymptotic to a single multiplicity @xmath5 plane .
in particular , when combined with a result of schoen @xcite , this resolves meeks full conjecture in the case of finite genus - that is , the only minimal surface with the area growth of two planes and finite genus is the catenoid . in this paper , we prove that meeks conjecture holds true under additional assumptions : [ thm : sublin1 ] let @xmath1 be an embedded minimal surface with the area growth of @xmath5 planes .
suppose that there exists @xmath11 such that for all @xmath12 sufficiently large , there exists a line @xmath13 @xmath14 is a union of at least @xmath15 disks @xmath16 and such that @xmath17 is homotopically nontrivial in @xmath18 .
then @xmath1 has a unique tangent cone at infinity .
this leads to the following : [ cor : sublin ] let @xmath1 be an embedded minimal surface with quadratic area growth .
let @xmath19 then if for some @xmath20 , @xmath21 is a union of @xmath15 topological disks @xmath16 each with finitely many boundary components , then @xmath1 has a unique tangent cone at infinity . note that the corollary substitutes the homotopy requirement from the theorem for the existence of a single line around which we can base our sublinearly growing set .
the author would like to thank his advisor , william minicozzi , as well as jonathan zhu , frank morgan , ao sun , and nick strehlke for their comments and suggestions throughout the writing of this paper .
the proof of this begins with the following : [ lem : lab ] suppose that @xmath1 satisfies the conditions of theorem [ thm : sublin1 ] .
then for some @xmath22 @xmath23 we will work on each leaf @xmath16 separately , and the lemma will come from adding the area of all the leaves together . first note that @xmath24 is a rotationally symmetric solid torus and ( since @xmath16 is a disk ) , @xmath17 is contractible in @xmath25 .
however , since @xmath25 is rotationally symmetric , the smallest spanning disk for any such curve has area at least that of a vertical cross section @xmath26 .
any such vertical cross section consists of a half - circle of radius @xmath12 minus a strip of length @xmath27 and width @xmath28 .
thus , we have @xmath29 note that lemma [ lem : lab ] implies that there are in fact exactly @xmath15 disks in the statement of theorem [ thm : sublin1 ] .
we make a definition : the * error * at scale @xmath30 of a minimal surfaces with area growth of @xmath5 planes is defined as @xmath31 thus , the lemma [ lem : lab ] is equivalent to the statement : @xmath32 we now apply an argument of brian white @xcite to prove uniqueness of the tangent cone . let @xmath1 satisfy the following : @xmath33 such that for @xmath34 , @xmath35 then @xmath1 has a unique tangent cone at infinity .
define @xmath36 . then note that @xmath37 is equal to the area of the projection of @xmath38 onto the unit sphere .
we will bound this area .
we have : @xmath39^{1/2 } \left[\int_{\sigma\cap b_r{\backslash}b_s}\frac{1}{|x|^2 } d\sigma\right]^{1/2 } \\
\end{aligned}\ ] ] by monotonicity , the term inside the first bracket is smaller than @xmath40 . also , the term in the second bracket can be bounded by distance and area .
thus , we get that the above is smaller than @xmath41 now , by equation ( [ eqn : errbd ] ) , along with the fact that @xmath42 , we have that this is bounded by @xmath43^{1/2 } \leq c\frac{r}{s^{(1-\alpha)/2 + 1}}\ ] ] pick @xmath44 and @xmath30 such that @xmath45 . then @xmath46 we then sum the above bound to see @xmath47 as @xmath48 , this term goes to zero .
thus , the area of the projection of @xmath49 approaches zero as @xmath30 gets large , which means that the tangent cone must be unique .
let @xmath16 be one of the components of @xmath50 . then note that the closure of @xmath16 in @xmath0 must be conformally equivalent to @xmath51 with finitely many boundary points removed .
take a neighborhood @xmath52 of one of these missing boundary points which does not come close to any other missing boundary points .
then @xmath53 has exactly one boundary component .
there are two options for the shape of @xmath54 . 1 .
the function @xmath55 is unbounded in both directions .
@xmath55 is bounded in one direction .
note that @xmath56 can not be bounded in both directions , as then @xmath54 would be compact , which it is not .
we temporarily assume that option 1 occurs .
let @xmath57 be the portion of @xmath54 which is not on the boundary of @xmath58 .
take @xmath20 larger so that @xmath59 .
let @xmath60 .
then some component of @xmath61 will satisfy the conditions of theorem [ thm : sublin1 ] .
this implies that it is possible to prove the lower area bound lemma for this component , and in particular , the area must be asymptotic to @xmath62 .
the following lemma will complete our proof : under our assumptions , option 2 is not possible .
suppose that option 2 occurs .
wlog , let @xmath55 be bounded below by 0 , and let @xmath63 be the point at which that minimum is achieved .
let @xmath64 .
let @xmath26 be a catenoid where the radius of the center geodesic is strictly larger than @xmath65 .
then by a simple application of the maximum principle , @xmath52 must intersect @xmath26 .
in particular , this implies that @xmath66 .
now , consider a sequence of @xmath67 such that @xmath68 converges to a tangent cone at infinity . by compactness , @xmath69 must either converge to a union of geodesics on @xmath70 or must disappear at infinity .
however , due to the discussion of the previous paragraph , @xmath52 can not disappear at infinty , and so must converge to a nontrivial union of geodesics @xmath71 , possibly with endpoints at the north or south poles .
let @xmath72 be a nonsmooth point on @xmath73 .
then there must exist a neighborhood @xmath74 of @xmath72 such that @xmath75 restricted to @xmath76 is unbounded as @xmath77 .
however , since @xmath52 is a minimal disk with quadratic area growth bounds , @xmath78 must be bounded by @xmath79 , where @xmath80 is the distance of @xmath81 from the boundary of @xmath52
. suppose that @xmath72 is not equal to the south pole .
then we can choose our neighborhood @xmath74 of @xmath72 to stay away from the @xmath56 axis , so we will have that @xmath82 uniformly on @xmath76 .
suppose that @xmath72 is equal to the south pole . then by the assumption of option 2 , @xmath54
is only contained in the region @xmath83 .
so , we can choose @xmath84 , and this implies the same uniform @xmath75 bound .
therefore , there will be no nonsmooth points of @xmath85 , which implies that @xmath71 consists of a single great circle passing through the north pole .
in particular , this implies that there are some @xmath86 such that the area of @xmath87 is greater than @xmath88 , where @xmath89 as @xmath90 .
thus , we have at least @xmath15 components of @xmath91 , each of which has area growth at least @xmath62 by the discussion of option 1
. however , since the global area growth is @xmath92 , no component can have growth @xmath93 .
there are several potential extensions of the work above . theorem
[ thm : sublin1 ] and corollary [ cor : sublin ] effectively assume that all tangent cones of @xmath1 are unions of planes _ with a common axis_. it is likely not significantly more difficult to show that the same result holds in the case when the one - dimensional singular set is more complicated , as long as away from a sublinearly growing neighborhood , @xmath1 is a union of disks .
that is , we have the following as another potential step towards the resolution of meeks conjecture : let @xmath1 have the area growth of @xmath5 planes , and suppose that there exists a uniform @xmath94 such that for each @xmath95 , the following is true : there exist line segments @xmath96 , @xmath97 such that outside of an @xmath98sublinearly growing neighborhood of @xmath99 , @xmath100 is a union of disks .
then @xmath1 has a unique tangent cone at infinity . | we partially resolve a conjecture of meeks on the asymptotic behavior of minimal surfaces in @xmath0 with quadratic area growth . |
spectral properties , such as the one - electron spectral function , provide valuable insight into the usually complex physics of strongly correlated systems . however , reliable results for such quantities are difficult to obtain .
analytical methods are often restricted to very simple limiting cases , and results can usually not be extended to more general situations .
two remarkable exceptions are the holstein model with linear electron dispersion@xcite and the hubbard model,@xcite both in one dimension , which have been solved exactly . to study
more general models , numerical methods such as exact diagonalization ( ed ) and quantum monte carlo ( qmc ) have received much attention over the last decades .
ed methods allow very accurate calculations of ground - state as well as finite - temperature spectral properties , but are restricted to rather small clusters due to the large dimension of the corresponding hilbert space .
qmc can be used to obtain results on large clusters even in higher dimensions , but here the sign - problem and the ill - posed analytic continuation required to obtain dynamic correlation functions for real frequencies are detrimental for many interesting applications .
the recently developed cluster perturbation theory@xcite ( cpt ) marks an important improvement of the situation .
it is based on a break - up of the infinite system into identical , finite clusters , on which the one - electron green function is calculated exactly . then , the hopping between clusters is treated within strong - coupling perturbation theory@xcite ( scpt ) .
the method has been successfully used for various hubbard models , to calculate spectral functions , as well as other quantities of interest , both for zero@xcite and finite temperature.@xcite although the concept of cpt relies on a model with local interactions only , it has also been applied with some success to the @xmath0 model.@xcite for the calculation of the cluster green function , the ed lanczos method ( for a review see ref . ) can be used .
for the case of coupled electron - phonon systems , such as the holstein model and its various extensionse.g.the holstein - hubbard or the holstein double - exchange model the application of ed methods is hampered by the infinite number of possible phonon configurations , which gives rise to a rapidly growing requirement of computer memory and/or cpu time as the number of lattice sites or phonon states increases .
consequently , standard ed methods employing some kind of hilbert space truncation are restricted to very small clusters , especially for low phonon frequency and/or strong electron - phonon coupling .
again improved methods such as dmrg , or the use of variational phonon bases allow to extend the accessible parameter range .
nevertheless , as electron - phonon interaction has been identified as an important ingredient in , e.g. , high - temperature superconductors@xcite and manganites@xcite , further progress along these lines is highly desirable . in this paper , we show that cpt can be successfully applied to electron - phonon models with a ( local ) coupling of the holstein type.@xcite we present results for the one - electron spectral function of the holstein polaron , i.e.the holstein model with one electron , in one and two dimensions . the holstein polaron problem has been investigated intensively in the past , and a wealth of information about its spectral properties is available .
we find that the use of cpt strongly reduces finite - size effects , giving results which are much closer to the thermodynamic limit than the corresponding ed data .
additionally , we consider the special case of a completely saturated ferromagnetic state at zero temperature in the holstein double - exchange model@xcite for colossal magnetoresistive manganites .
the latter is then equivalent to the holstein model of spinless fermions , and we combine the exact atomic - limit one - particle green function with cpt for a single - site cluster to calculate the spectral function at quarter - filling .
the results of this simple approach agree well with the previously developed many - body coherent potential approximation.@xcite the paper is organized as follows : in sec .
[ sec : cpt ] we give a review of cpt .
[ sec : holsteinpolaron ] discusses the application to the holstein polaron , while the scpt for the many - electron case is presented in sec .
[ sec : holsteinde ] . finally , sec .
[ sec : summary ] contains our conclusions .
the basic idea of cpt@xcite is to divide the infinite lattice into identical clusters , each containing @xmath1 lattice sites . adopting the notation of ref .
, the hamiltonian of the system is written in the form @xmath2 where @xmath3 in the most general formulation of cpt,@xcite the subscripts @xmath4 , @xmath5 denote different orbitals within a cluster , but here we restrict ourselves to the case of one orbital per site so that @xmath4 , @xmath6 . the vectors @xmath7 , @xmath8 correspond to sites in the superlattice of clusters ( see ref . ) . in eq .
( [ eq : h0pv ] ) , @xmath9 represents a hamiltonian of a single cluster containing local interactions only and @xmath10 describes the hopping between clusters , i.e.the hopping amplitude between site @xmath4 of cluster @xmath7 and site @xmath5 of cluster @xmath8 is given by the matrix element @xmath11 .
although long - range hopping can also be included,@xcite we shall only consider models with nearest - neighbor hopping so that @xmath12 for neighboring sites @xmath4 , @xmath5 in adjacent clusters @xmath7 , @xmath8 . within cpt , an approximation for the green function of the original system , @xmath13 , is obtained using an analytical strong - coupling perturbation expansion up to first order of the inter - cluster hopping @xmath10 ( for details of the derivation see ref . ) .
the resulting equation relating the green function of the original lattice to the energy - dependent cluster green function @xmath14 reads@xcite @xmath15 here @xmath16 , @xmath14 and @xmath17 stand for @xmath18 matrices , and the inter - cluster hopping @xmath10 has been partially fourier - transformed exploiting the translational symmetry of the cluster superlattice , with @xmath19 being a wave vector of the reduced brillouin zone.@xcite finally , the green function @xmath20 can be transformed from the mixed representation of eq .
( [ eq : cpt1 ] ) , real space within a cluster and reciprocal space between clusters , using@xcite @xmath21 to obtain the familiar representation of the one - electron green function . @xmath22 as given by eqs .
( [ eq : cpt1 ] ) and ( [ eq : cpt2 ] ) becomes exact in the atomic limit @xmath23 ( refs . ) .
moreover , it also reduces to the exact result for the case of non - interacting electrons@xcite since , in this case , eq .
( [ eq : cpt1 ] ) corresponds to the exact resummation of the perturbation series .
finally , cpt also becomes exact in the limit @xmath24 ( refs . ) .
the one - electron cluster green function at zero temperature @xmath25 can be calculated exactly for any pair of site indices @xmath4 , @xmath5 in the cluster using , e.g. , the lanczos method . here
@xmath26 is the energy of the ground state @xmath27 of the cluster , and a spin index has been suppressed in the notation .
the two parts of the green function matrix @xmath28 correspond to adding or removing an electron to / from @xmath27 .
finally , the one - electron spectral function is defined as @xmath29 since cpt is based on a perturbation expansion in the inter - cluster hopping , the method can be expected to work especially well in the strong - coupling regime .
this is also illustrated by the fact that it becomes exact in the atomic limit , as mentioned above .
on the other hand , for weak or intermediate coupling , the electronic kinetic energy is not small compared to the local interactions .
consequently , the size of the cluster has to be large enough in order to obtain accurate results .
in fact , from previous applications of cpt , e.g. , to the one- and two - dimensional hubbard model,@xcite the cluster size @xmath1 emerged as the main control parameter of the method . in the case of the one - dimensional hubbard model , for example
, @xmath30 is identical to the hubbard i approximation,@xcite while @xmath31 already gives a spectral function that contains most of the relevant features such as short - range antiferromagnetic ordering.@xcite with increasing @xmath1 , the cpt green function approaches systematically the exact result for the infinite system . for identical cluster size
, the cpt spectrum contains many more poles with significant residues than the corresponding results of ed .
in fact , also in the 1d hubbard model , the spectrum obtained with cpt on a four - site cluster is already comparable in quality to the ed spectrum for @xmath32 ( refs . ) .
an additional advantage of cpt is the possibility to evaluate @xmath33 at continuous wavevectors @xmath34 , in contrast to ed which restricts @xmath34 to the @xmath1 vectors of the first brillouin zone of the cluster , of which only @xmath35 are physically distinct .
finally , finite temperature lanczos methods can also be combined with cpt to calculate thermodynamic properties.@xcite concerning the application of the lanczos method to calculate the cluster green function @xmath28 , it is important to stress the need for open boundary conditions ( bcs ) .
attempts have been made to use periodic bcs and subtract the corresponding terms afterward in the perturbative treatment of the inter - cluster hopping , but it has been found that the accuracy of the results is much better for the case of open bcs . although the latter are physically more intuitive in connection with cpt , the calculation of the cluster green function with the lanczos method becomes more difficult as one can not exploit translational symmetry .
other symmetries such as the inversion group can be used in principle , but are usually not as effective in saving computer memory by reducing the size of the corresponding hamiltonian matrix and lanczos vectors .
we want to point out that cpt does not in principle rely on the ed method .
in fact , the cluster green function may be calculated using any method available.@xcite indeed we will see in sec . [
sec : holsteinde ] that it is possible to combine the exact analytic solution for the atomic - limit green function with cpt , to obtain results which agree surprisingly well with the many - body coherent potential approximation.@xcite in addition to the spectral function considered here , other physical properties of the system can also be calculated with cpt .
this includes , e.g. , the ground state energy of the infinite system , the electronic kinetic energy or the fermi surface.@xcite the strength of cpt lies in the calculation of the one - particle green function and related quantities such as the density of states .
the numerical effort is relatively small compared to more sophisticated methods like dmrg or qmc .
an additional advantage of cpt is the fact that it can easily be applied also to two - dimensional systems , in contrast to , e.g. , dmrg .
finally , an important disadvantage of cpt should be mentioned : within the current formulation , two - particle green functions can not be calculated .
consequently , it is not possible to compute , e.g. , the dc conductivity or other interesting two - particle correlation functions .
the hamiltonian of the holstein model@xcite reads @xmath36 where @xmath37 ( @xmath38 ) and @xmath39 ( @xmath40 ) are creation ( annihilation ) operators for an electron with spin @xmath41 and a phonon of frequency @xmath42 at lattice site @xmath43 , respectively .
the electron occupation number is defined as @xmath44 with @xmath45 , and the parameters of the model are the hopping integral for nearest - neighbor hopping , @xmath46 , and the electron - phonon coupling strength , denoted as @xmath47 .
it is common to define a dimensionless coupling parameter @xmath48 , where @xmath49 is the bandwidth of the bare electron band , and a dimensionless phonon frequency @xmath50 .
the holstein model can then be described using only these two parameters .
moreover , we shall express all energies in units of @xmath46 . as mentioned above ,
here we only consider the one - electron limit of hamiltonian ( [ eq : hamilt ] ) which is also called the holstein polaron problem .
although there is only a single electron in the system , the coupling to the phonons makes it a complex many - body problem , which has been the focus of much theoretical work .
the restriction to one electron greatly simplifies calculations with the lanczos method since both , the number of required phonon states@xcite and the number of electron configurations grow noticeably with the number of particles .
however , in sec .
[ sec : holsteinde ] , we will use the exact result for the atomic - limit green function and cpt for a single - site cluster to calculate the spectral function of the holstein model of spinless fermions at quarter - filling .
following other authors,@xcite we calculate the green function @xmath51 where @xmath52 represents the ground state of the phonons and the vacuum state for the electrons .
the spin index can be suppressed due to the symmetry of the problem .
the corresponding one - electron spectral function is given by eq .
( [ eq : akw ] ) . compared to the class of hubbard models for which cpt has been originally developed
, we are facing an additional difficulty arising from the priori infinite number of allowed phonon states .
we employ a widely used truncation scheme@xcite of the phonon hilbert space which is spanned by the basis states @xmath53 where @xmath54 denotes the number of phonons at lattice site @xmath43 .
now the truncation consists in restricting the basis states to the subset with @xmath55 leading to @xmath56 allowed phonon configurations .
the convergence of the results with @xmath57 can be monitored using the ground - state energy @xmath58 of the cluster with open bcs . in all results of this paper , @xmath57 was chosen such that the relative error for the ground state with one electron , @xmath59 , was smaller than @xmath60 .
we find that convergence of @xmath58 also ensures a well - converged spectral function .
moreover , the influence of the number of phonons kept in the calculation is much larger for the incoherent part of the spectrum than for the coherent , low - energy quasi - particle peak which determines @xmath58 ( see sec . [
sec : comp - with - exact ] ) .
finally , a refined truncation scheme which allows for extremely accurate results ( relative error @xmath61 ) has been proposed by wellein et al.@xcite .
before we come to a discussion of the results obtained with cpt , we want to comment on some of the existing work on spectral properties of the holstein polaron .
as indicated before , the most reliable method to calculate dynamic quantities , such as @xmath33 , is ed which has been used extensively in the past .
@xcite most of this work has focused on the polaron band structure @xmath62 instead of the spectral function , since it is often easier to interpret , especially in the strong - coupling regime where the structure of @xmath33 is rather complicated . however , as pointed out by wellein et al.,@xcite the two quantities are closely related .
in fact the position of the lowest - energy peak in @xmath33 , obtained from the green function ( [ eq : g_lanczos ] ) , follows exactly the polaron band structure as we vary @xmath34 .
moreover , as discussed by wellein and fehske,@xcite the integral over this first peak is equivalent to the quasi - particle ( qp ) weight @xmath63 , where @xmath64 denotes the lowest - energy single - polaron state in the sector with total momentum @xmath34 .
other numerical methods which have been used to calculate the spectrum of the holstein polaron include dmrg@xcite ( in one dimension ) , finite - cluster strong coupling perturbation theory@xcite
( 1d , 2d ) , qmc@xcite ( 1d3d ) , and variational methods@xcite ( 1d4d ) . as mentioned above
, the critical parameter of cpt is the number of sites in the cluster . to demonstrate the advantage of cpt over the standard ed method ( see , e.g. , ref . )
we present in fig .
[ fig : lan_cpt_weak ] the spectral function @xmath65 in one dimension for different cluster sizes @xmath1 .
we chose @xmath66 and @xmath67 , which is the regime where an extended polaron exists ( see , e.g. , ref . ) . consequently
, significant finite - size effects can be expected for small clusters , which is exactly what we see in the ed results . for the latter periodic bcs
have been used .
[ fig : lan_cpt_weak ] clearly shows that the shape of the large qp peak at @xmath68 changes very little with increasing @xmath1 for both , ed and cpt , but a noticeable shift can be observed in the case of the ed spectra as we go from @xmath31 to @xmath69 .
the influence of @xmath1 is much larger for the incoherent part of the spectrum , which lies about a distance @xmath70 above the qp peak .
the ed spectra display sharp , well - separated peaks , whereas the corresponding cpt data containing many more poles resembles much closer the expected results for an infinite system .
the latter has been investigated by marsiglio@xcite using migdal - eliashberg theory .
for the same parameters , he found that the qp peak remains almost unchanged as @xmath24 , while the incoherent part evolves into a continuous band that fits well to the cpt results even for rather small clusters @xmath71 .
we have also compared @xmath72 for @xmath73 , and the observed influence of finite - size effects agrees perfectly with previous work of wellein and fehske:@xcite as @xmath74 increases from @xmath75 to @xmath76 , the size of the polaron increases , and the deviations of the ed data from the cpt results become larger . in the strong - coupling or small - polaron regime , not shown here
, finite - size effects are known to be small .
consequently , even for very small clusters , ed and cpt both give well - converged results for the qp peak which determines , e.g. , the ground - state energy .
however , in the case of ed , the incoherent part of the spectrum for wavevector @xmath74 , corresponding to excitations of an electron with momentum @xmath77 and a phonon with momentum @xmath78 , still exhibits the typical multi - peak structure of a finite system , whereas the cpt results again reproduce much better the incoherent band found in the thermodynamic limit .
moreover , as mentioned in sec .
[ sec : cpt ] , cpt allows to calculate @xmath72 for continuous @xmath74 , while ed on a @xmath1-site cluster is restricted to @xmath35 physically non - equivalent wavevectors .
comparison of the spectral function @xmath65 of the 1d holstein polaron obtained with ed ( left column ) and cpt ( right column ) for various numbers of lattice sites @xmath1 in the cluster .
the plot is for @xmath79 , @xmath67 and @xmath80 .
an artificial imaginary part @xmath81 has been used to broaden the delta peaks.,scaledwidth=45.0% ] a closer look at the cpt results in fig .
[ fig : lan_cpt_weak ] reveals small additional peaks not present in the ed spectra which move from the incoherent part of @xmath72 towards the qp peak with increasing @xmath1 .
additional calculations for larger clusters have shown that these peaks vanish systematically with increasing @xmath1 , so that the cpt spectrum approaches the exact result in the thermodynamic limit @xmath82 , as expected .
consequently , these peaks are not a defect of cpt , but represent finite - size effects which arise from the approximate treatment of inter - cluster hopping .
the latter , in combination with the open bcs used to calculate the cluster green function , leads to a system which does not have perfect translational symmetry .
the situation is equivalent to ed with open bcs : for @xmath24 the spectrum approaches the results of an infinite cluster .
however , in contrast to cpt , the effects for finite @xmath1 are much more significant . moreover , these finite - size effects manifest themselves in a slightly different way than in the case of periodic bcs , where no additional peaks showing the afore mentioned behavior are found . in the case of cpt , already for the small cluster sizes shown in fig . [
fig : lan_cpt_weak ] , the spectral weight of these additional peaks is extremely small compared to the rest of the spectrum . for other values of @xmath70 and @xmath83 ,
a similar behavior has been found .
although not discussed by the authors , similar effects can also be expected for the case of the hubbard model,@xcite although they may be larger for the holstein polaron due to the higher sensitivity of phonon - excitations to the bcs .
top : spectral function @xmath72 of the 1d holstein polaron calculated with cpt for @xmath84 , @xmath80 and @xmath81 .
bottom : density plot of the same data for 100 points in @xmath74 space .
symbols represent results of bona et al.@xcite , title="fig:",scaledwidth=41.0% ] + top : spectral function @xmath72 of the 1d holstein polaron calculated with cpt for @xmath84 , @xmath80 and @xmath81 .
bottom : density plot of the same data for 100 points in @xmath74 space .
symbols represent results of bona et al.@xcite , title="fig:",scaledwidth=45.0%,scaledwidth=30.0% ] in one dimension , the general picture emerging from previous work on the holstein polaron problem is as follows ( see e.g.ref . and references therein )
: in the non - adiabatic regime ( @xmath85 ) , a so - called lang - firsov polaron is formed which , due to the instant response of the phonons to the electronic motion , represents a very localized object . as the electron - phonon coupling increases , its mobility or effective hopping amplitude exhibits a gradual decrease , and for strong coupling a nearly - localized small polaron moving in an exponentially narrow band exists .
in contrast , in the adiabatic ( @xmath86 ) , weak - coupling regime the electron drags with it an extended cloud of phonons .
this object is usually called a ` large polaron ' . at @xmath87 , a sharp but continuous@xcite transition to
a less mobile small holstein polaron takes place .
the two conditions for small - polaron formation , independent of the value of @xmath70 , are @xmath88 _ and _ @xmath89.@xcite moreover , as pointed out by capone et al.,@xcite the formation of small polarons is determined by @xmath88 for @xmath86 and by @xmath89 for @xmath85 . here
we restrict ourselves to the most interesting regime of phonon energies comparable to the electronic hopping , i.e.@xmath90 . for intermediate @xmath70 and @xmath83 ,
no reliable analytical methods exist , so that numerical approaches represent the most important source of information . in fig .
[ fig : spectrum1d_lambda0.25 ] we present results for @xmath72 for @xmath91 , @xmath92 and @xmath84 , as well as a density plot of the same data . as mentioned before , the spectrum consists of a low - lying qp peak and an incoherent part at higher energies .
the physics behind the observed behavior of @xmath72 has been discussed , e.g. , by stephan,@xcite and is typical for electronic systems weakly interacting with dispersionless optical phonons . for small @xmath74 ,
most of the spectral weight resides in the qp peak which corresponds to a weakly - dressed electron . for the case considered here ,
in which the phonon energy lies inside the bare electron band , electron and phonon hybridize and repel each other near the point where they would be degenerate , i.e.for @xmath93 .
this coincides with the region where the flattening of the polaron band occurs and , in fact , for larger @xmath74 the phonon becomes the lowest - energy excitation .
however , most of the spectral weight is contained in the broad , incoherent band which follows the free - electron dispersion . the density plot in fig .
[ fig : spectrum1d_lambda0.25 ] also contains data for the polaron band structure @xmath94 which has been obtained by bona et al.@xcite using their variational ed method .
the latter has been shown to give very accurate results for the infinite system , although it becomes somewhat less accurate in the strong coupling regime and for large values of @xmath74 ( ref . ) . as mentioned before
, @xmath94 corresponds to the lowest - energy band in @xmath72 and we find very good agreement with our data throughout the brillouin zone . fig .
[ fig : spectrum1d_lambda0.5 ] shows results for a similar phonon frequency @xmath95 but for stronger electron - phonon coupling @xmath67 and @xmath32 .
compared to the weak - coupling case discussed above , the polaron band is separated more clearly from the incoherent part of the spectrum and , as expected , the band - width is further reduced . additionally ,
even more spectral weight has been transfered to the high - energy , incoherent band . on top of that
, a gap shows up in the upper band at about @xmath96 .
again the polaron band fits very well the results for @xmath94 of bona et al.@xcite we next consider the case of intermediate coupling @xmath97 , with @xmath98 and @xmath99 ( fig .
[ fig : spectrum1d_lambda1.0 ] ) .
for these parameters , an extended polaron exists which still has a relatively large band - width , compared to the small - polaron case discussed below .
moreover , the incoherent part of the spectrum has split up into several sub - bands separated in energy by @xmath70 , which correspond to excitations of an electron and one or more phonons . as
before , we find very good agreement between the low - energy band in @xmath72 and the polaron band energy @xmath94 calculated by bona.@xcite finally , in fig .
[ fig : spectrum1d_lambda2.0 ] , we report the spectral function for @xmath98 and @xmath100 .
the results have been obtained using only a four - site cluster , which is sufficient to get very good agreement with bona s data for @xmath94 , with only minor deviations at large values of @xmath74 where finite - size effects are most pronounced , as discussed in sec .
[ sec : comp - with - exact ] .
this is a consequence of the predominantly local effects in the strong - coupling regime , which also manifest themselves in terms of a very narrow polaron band .
top : spectral function @xmath72 of the 1d holstein polaron calculated with cpt for @xmath32 , @xmath80 and @xmath81 .
bottom : density plot of the same data for 100 points in @xmath74 space .
symbols represent results of bona et al.@xcite , title="fig:",scaledwidth=41.0% ] + top : spectral function @xmath72 of the 1d holstein polaron calculated with cpt for @xmath32 , @xmath80 and @xmath81 .
bottom : density plot of the same data for 100 points in @xmath74 space .
symbols represent results of bona et al.@xcite , title="fig:",scaledwidth=45.0%,scaledwidth=30.0% ] top : spectral function @xmath72 of the 1d holstein polaron calculated with cpt for @xmath99 , @xmath101 and @xmath81 .
bottom : density plot of the same data for @xmath102 points in @xmath74 space .
symbols represent results of bona.@xcite , title="fig:",scaledwidth=41.0% ] + top : spectral function @xmath72 of the 1d holstein polaron calculated with cpt for @xmath99 , @xmath101 and @xmath81 .
bottom : density plot of the same data for @xmath102 points in @xmath74 space .
symbols represent results of bona.@xcite , title="fig:",scaledwidth=45.0%,scaledwidth=30.0% ] top : spectral function @xmath72 of the 1d holstein polaron calculated with cpt for @xmath69 , @xmath103 and @xmath81 .
bottom : density plot of the same data for @xmath102 points in @xmath74 space .
symbols represent results of bona.@xcite , title="fig:",scaledwidth=41.0% ] + top : spectral function @xmath72 of the 1d holstein polaron calculated with cpt for @xmath69 , @xmath103 and @xmath81 .
bottom : density plot of the same data for @xmath102 points in @xmath74 space .
symbols represent results of bona.@xcite , title="fig:",scaledwidth=45.0%,scaledwidth=30.0% ] to illustrate the applicability of cpt , we also calculated the spectral function of the holstein polaron on a 2d cluster with @xmath99 , which has the shape of a tilted square .
in contrast to cpt in one dimension , the choice for the shape of the cluster is not unique , and different clusters may lead to slightly different results .
this possibility has been investigated for the hubbard model , and the effect of the cluster shape on @xmath33 was found to be rather small.@xcite for the case of the holstein polaron , where the physics is dominated by local correlations , the influence of the geometry of the cluster is expected to be even smaller .
as discussed , e.g. , by wellein et al.,@xcite similar to 1d , a small polaron is formed in two dimensions provided that @xmath88 and @xmath89 . while the behavior in the non - adiabatic regime ( @xmath85 ) is only weakly affected by dimensionality,@xcite
important differences exist in the adiabatic regime @xmath86 : in contrast to the one - dimensional case , where a large polaron is formed for any @xmath104 , the electron remains quasi - free for @xmath105 , as indicated by an almost unaffected effective hopping amplitude .
moreover , for the same value of @xmath70 , the cross - over to a small polaron at @xmath87 is much sharper in 2d than in 1d .
here we simply aim to demonstrate the possibility of calculating the 2d spectral function with cpt .
therefore , we restrict ourselves to one set of parameters , namely @xmath79 and @xmath106 , which has also been treated using _ finite - cluster _ strong - coupling perturbation theory@xcite .
in contrast to standard scpt based on the lang - firsov transformation , the latter has been shown to give reliable results also for intermediate @xmath83 and @xmath70 , which is a consequence of the inclusion of longer - ranged effects.@xcite while in the 1d case the density plot of @xmath33 contains all 100 values of @xmath74 used in cpt , in two dimensions we have used 400 points in @xmath34 space .
however only 60 , lying along @xmath107 , are shown in fig .
[ fig : spectrum2d ] . from the above discussion , and for the parameters considered here
, we expect a rather broad polaron band .
this is clearly confirmed by the spectral function shown in fig .
[ fig : spectrum2d ] , and the lowest - energy band in our data resembles closely the findings of stephan ( fig . 2 of ref . ) .
in particular , similar to the one - dimensional case considered in sec .
[ sec:1d - holstein - polaron ] , a flattening of the polaron band near @xmath108 is found which has also been noted by wellein et al.@xcite above the polaron band , also similar to 1d , there lie several other incoherent bands which correspond to multi - phonon excitations and are therefore separated in energy by @xmath70 .
top : spectral function @xmath33 of the 2d holstein polaron calculated with cpt for @xmath99 , @xmath101 and @xmath81 .
bottom : density plot of the same data ( see text).,title="fig:",scaledwidth=41.0% ] + top : spectral function @xmath33 of the 2d holstein polaron calculated with cpt for @xmath99 , @xmath101 and @xmath81 .
bottom : density plot of the same data ( see text).,title="fig:",scaledwidth=45.0%,scaledwidth=30.0% ] in summary , the results of this section clearly demonstrate that cpt is applicable not only in the strong - coupling regime , but also for weak and intermediate electron - phonon interaction .
the quality of the resulting spectra is superior to ed data for the same cluster size , and very good agreement has been found with the variational method of bona et al.@xcite in one dimension . moreover , we have shown
that cpt also allows accurate calculations of @xmath33 in two dimensions .
in the last section , we have restricted ourselves to the holstein model with one electron .
although cpt has been successfully applied , e.g. , to the many - electron hubbard model,@xcite the electron - phonon coupling in the holstein model greatly complicates calculations using lanczos ed . for finite electron density , we combine cpt with the exact analytic result for the green function in the atomic limit . for the atomic limit , the green function has been obtained for many models using the equation - of - motion method,@xcite and here it will allow us to obtain results for the many - electron case which will be compared with the many - body coherent potential approximation discussed below . extending previous work of edwards et al.@xcite for the pure double - exchange ( de ) model ( see , e.g. , ref . ) , green@xcite studied the holstein - de model using a many - body coherent potential approximation ( cpa ) which , owing to the more complicated form of the holstein - de hamiltonian , constitutes a considerable extension of the hubbard iii approximation.@xcite the many - body cpa successfully describes many aspects of the manganites , and we refer the reader to a recent review of this work by edwards.@xcite here we only consider the special case of a completely saturated ferromagnetic state at temperature @xmath109 , with all itinerant spins having @xmath110 spin , say .
consequently , the de term which couples local and itinerant spins@xcite becomes merely a constant shift in energy , and the holstein - de model is equivalent to the pure holstein model of spinless fermions , i.e.with no doubly - occupied sites.@xcite an important feature of the many - body cpa is that the one - electron green function reduces to the exact atomic limit for @xmath23 , which takes the form@xcite @xmath111 where @xmath112 and the polaron binding energy @xmath113 ( @xmath114 , 1 ) has been absorbed into the chemical potential .
the general result for @xmath115 of the holstein model with electrons of both spins has been given by green,@xcite and we drop the spin - index in the sequel . as discussed by edwards,@xcite for an elliptic density of states , the local green function @xmath116 for complex energy @xmath117 satisfies the cpa equation @xmath118 and the self - energy can be obtained from@xcite @xmath119 finally , the one - electron spectral function is given by @xmath120\,,\ ] ] where @xmath121 is the band energy for wavevector @xmath34 . in order to compare with angle - resolved photoemission ( arpes ) data on the bilayer manganite @xmath122 , nominally with @xmath123 ,
hohenadler and edwards chose a strong - electron phonon coupling @xmath124 , as deduced from the low curie temperature of this material.@xcite to simplify calculations , they also used @xmath125 for which case the chemical potential @xmath126 by symmetry . we want to point out that the many - body cpa assumes a homogeneous system , so that no tendencies toward charge - density - wave order occur as @xmath127 is varied.@xcite as in previous work,@xcite hohenadler and edwards used @xmath128ev and @xmath129 ( see also ref . ) .
the results@xcite for @xmath33 , shown in fig . [ fig : cpa - cpt ] , support the theory of alexandrov and bratkovsky@xcite that in these manganites , small polarons exist in the ferromagnetic state . a similar interpretation of the experimental data based on standard small - polaron theory
had also been given by dessau et al.@xcite well away from the fermi surface , a well - defined peak exists which broadens as @xmath34 approaches the fermi level @xmath130 at @xmath131 . if @xmath132 is increased further , most of the spectral weight is transfered above @xmath130 . moreover
, the peaks never approach the fermi level closely , in agreement with the experimental data .
this indicates the existence of a pseudogap in the one - electron density of states .
however , in the gap , there exist small polaron sub - bands ( see fig . 4 of ref . ) and one of them , at the fermi level , presumably gives rise to the low but finite conductivity of the system .
as discussed by edwards,@xcite the many - body cpa does not give coherent states with infinite lifetime at the fermi level , even for @xmath109 .
this is typical for any cpa , and here it leads to an incoherent polaron sub - band around the fermi level . nevertheless , outside the central band around @xmath130 , the imaginary part of the self - energy displays the correct behavior , i.e.it vanishes in the gap , between the polaron bands .
comparison of the spectral function of the holstein model of spinless fermions at @xmath109 , calculated with scpt ( top ) and with the many - body cpa ( bottom , taken from ref . ) . here
the wavevector @xmath34 is given by @xmath133 with @xmath132 as indicated in the figure .
the plot is for @xmath129 and @xmath124 .
the scpt results have been broadened using a smearing parameter @xmath134.,title="fig:",scaledwidth=35.0% ] + comparison of the spectral function of the holstein model of spinless fermions at @xmath109 , calculated with scpt ( top ) and with the many - body cpa ( bottom , taken from ref . ) . here
the wavevector @xmath34 is given by @xmath133 with @xmath132 as indicated in the figure .
the plot is for @xmath129 and @xmath124 .
the scpt results have been broadened using a smearing parameter @xmath134.,title="fig:",scaledwidth=35.0% ] in this section we use the exact result for the atomic - limit green function of the holstein model of spinless fermions , @xmath115 ( eq . ( [ eq : g_al ] ) ) , and combine it with cpt to compare the resulting spectrum with the many - body cpa . for this case of a single - site cluster ( @xmath135 ) , eqs .
( [ eq : cpt1 ] ) and ( [ eq : cpt2 ] ) reduce to a single equation for the one - electron green function@xcite @xmath136 with @xmath137 and @xmath138 as defined by eq .
( [ eq : band ] ) . hence , as mentioned before , cpt for @xmath30 is equivalent to the hubbard i approximation,@xcite but here with the more complicated atomic - limit green function of the holstein model given by eq .
( [ eq : g_al ] ) . in the sequel
, we shall refer to this approximation as scpt .
this is justified by the fact that the approach becomes exact for @xmath23 . historically , a similar strong - coupling expansion for the hubbard model@xcite including higher order corrections has been the starting point for the development of cpt .
before we discuss the results , we would like to comment on the quality of the scpt used here : while the many - body cpa requires a self - consistent , iterative solution of eq .
( [ eq : cpa ] ) , the scpt green function is obtained from the lehmann representation of the atomic - limit green function ( [ eq : g_al ] ) , and the subsequent use of the resulting self - energy @xmath139 in eq .
( [ eq : gf_k ] ) .
similar to the original hubbard i approximation,@xcite the resulting green function consists of delta peaks corresponding to states with infinite lifetime .
however , due to the poles in the self - energy , there are no states at the fermi level and the system is not a fermi liquid .
as in the many - body cpa , @xmath140 depends on @xmath34 only through the band energy @xmath138 , whereas the self - energy is local .
this reliance on the atomic limit is reasonable in the strong - coupling regime considered here , where small polarons move in an extremely narrow band .
consequently , the simple perturbative treatment of the hopping term can be expected to give sensible results .
nevertheless , in scpt , we have to use an artificial imaginary part @xmath141which does not depend on energy to obtain peaks of finite width .
although for large enough @xmath141 there will be states at the fermi level , the latter have only finite lifetime even for @xmath109 .
hence , both scpt and the many - body cpa never give a fermi liquid , but the self - consistent cpa green function yields an imaginary part of the self - energy that shows the correct , strong energy - dependence except for the region inside the very small , incoherent polaron band around @xmath130 , as discussed in sec .
[ sec : many - body - cpa ] .
thus , as could be expected from the hubbard i - like approximation in eq .
( [ eq : gf_k ] ) , the many - body cpa is superior to scpt , although both approaches become exact in the atomic limit . the spectral function obtained with scpt using eq .
( [ eq : akw ] ) , also shown in fig .
[ fig : cpa - cpt ] , resembles quite closely the results of hohenadler and edwards.@xcite although there are some differences concerning the width and the position of the peaks , the overall behavior is very similar . in particular , the broadening of the qp peak near the fermi surface at @xmath131 is well reproduced .
clearly , the success of scpt consists in a surprisingly good agreement with the cpa data for all @xmath34 . despite this agreement
, cpt fails to reproduce the polaron sub - bands , and the sharp edge to the pseudogap for large values of @xmath132 .
moreover , the gap is larger than in the cpa data .
these shortcomings are a consequence of the rather crude approximation . nevertheless , keeping in mind the simplicity of the ansatz ,
the agreement with the many - body cpa is satisfactory .
we would like to point out that the scpt presented here can also be generalized to the holstein - de model with quantum spins ( e.g.@xmath142 appropriate for the manganites@xcite ) and at finite temperature , using the atomic - limit green function given by edwards.@xcite finally , the approximation could be systematically improved by increasing the number of sites in the cluster , which is exactly the idea behind cpt .
however , for @xmath143 the cluster green function can no longer be calculated analytically and one has to resort to numerical methods such as ed as in sec .
[ sec : holsteinpolaron ] .
such calculations are extremely difficult for the case of quarter - filled two- or three dimensional clusters , small phonon frequency and strong electron - phonon coupling .
future work along these lines employing optimized phonon approaches@xcite ( see also sec . [
sec : summary])is highly desirable in order to assess the quality of the many - body cpa results .
we have applied cluster perturbation theory to the holstein polaron problem in one and two dimensions , and comparison with existing work has revealed very good agreement . in combination with the lanczos method to calculate the cluster green function ,
the method gives reliable results for the one - electron spectral function @xmath33 , which become exact in the weak- and strong - coupling limit , @xmath144 and @xmath23 , respectively , and for the case of an infinite cluster .
calculations for continuous values of the wavevector @xmath34 are possible and , more importantly , finite - size effects are significantly reduced compared to standard ed .
our results extend previous applications of cpt to hubbard and @xmath0 models , showing that the method is also well - suited for electron - phonon models with local interactions .
as pointed out before , for more than one electron in the system , it becomes increasingly difficult to include enough phonon states so as to obtain converged results .
future work may therefore combine optimized phonon approaches@xcite and cpt to investigate more complicated problems such as the many - electron case , or extended models including , e.g. , a hubbard term .
the major advantage of such an approach is the possibility of using relatively small clusters while still obtaining results which are only weakly influenced by finite - size effects .
additionally , we have used the exact atomic - limit green function of the holstein model of spinless fermions , to calculate the spectral function for the case quarter - filling and strong electron - phonon coupling .
the results of this approximation at the hubbard i level agree surprisingly well with the many - body cpa .
m.h . and m.a .
were supported by doc [ doctoral scholarship program of the austrian academy of sciences ] .
we are grateful to janez bona for providing us with previously unpublished data , and we acknowledge helpful discussions with david snchal and maria daghofer . finally , we would like to thank david m. edwards for making valuable comments on the manuscript . | cluster perturbation theory in combination with the lanczos method is used to compute the one - electron spectral function of the holstein polaron in one and two dimensions .
it is shown that the method allows reliable calculations using relatively small clusters , and at the same time significantly reduces finite - size effects .
results are compared with exact data and the relation to existing work is discussed .
we also use a strong - coupling perturbation theory
equivalent to the hubbard i approximation to calculate the spectral function of the quarter - filled holstein model of spinless fermions , starting from the exact atomic - limit green function .
the results agree well with previous calculations within the many - body coherent potential approximation . |
the classification of low dimensional complex associative algebras is one of the earliest classification theorems , dating to 1870 .
the notion of deformations of associative algebras is more recent , and was thoroughly studied by murray gerstenhaber in the 1960 s .
recently , the notion of moduli spaces of algebras of fixed dimension has appeared in the literature , which gives a new , more geometric view of the space of equivalence classes of algebras .
the classification of the algebras refers only to the set of equivalence classes , and can be said to have been solved if representatives for all equivalence classes have been found and described in terms of some families and special elements .
the main concern in the classification problem is to avoid duplicate listings .
the moduli space problem is to give a description of these equivalence classes in terms of geometry .
the set of all associative algebra structures is described by some homogenous quadratic polynomials on the structure constants , so these algebras determine an algebraic set .
the group of automorphisms acts on the space of structure constants , and preserves this algebraic set .
the moduli space is the quotient by this group action , and thus is not a variety , and as a topological space , is not hausdorff . in this paper
, we give a different interpretation of this moduli space , by decomposing the set of equivalence classes of associative algebras according to deformation theory .
more specifically , the versal deformations of an element in this space give a picture of a local neighborhood of the element .
from this local analysis , the authors were able to determine that for lie algebras at least up to dimension 4 ( see @xcite ) , a geometric picture emerges .
even though the moduli space is not hausdorff , it has a stratification by orbifolds , which are connected by special types of deformations , called jump deformations , which capture all of the non - hausdorff behaviour of the moduli space .
our primary aim in this paper was to examine whether the same phenomena applied to moduli spaces of associative algebras .
as the results in this paper show , the moduli space of complex associative algebras of dimension 3 has a stratification of the same type observed for lie algebras .
there are parallels and differences between the moduli spaces of complex lie algebras and associative algebras on a space of dimension @xmath0 .
for example , for 2-dimensional spaces , all strata are singleton points , but in the case of associative algebras there are six such points , and there is only one complex lie algebra on a space of dimension 2 . in this paper
, we shall show that on a 3-dimensional complex space , the moduli space of complex associative algebras consists of 22 different strata , of which 21 are singleton points , and one an orbifold modelled on the space @xmath1 . for the 3-dimensional lie algebra case , there were 4 strata , 3 of which were singleton points , and one given by the orbifold @xmath1 .
complex associative algebras of dimension up to 5 were first classified by benjamin peirce as early as 1870 , originally in the form of a self - published text , which appeared later in @xcite .
peirce s methodology was criticized by his son charles peirce , who pointed out that the quaternions can not be realized as a complex associative algebra , but are only a real algebra .
it was charles peirce who first showed that the only division algebras over the reals are the real , complex and quaternionic algebras ( see @xcite ) . in @xcite , an analysis of peirce s method of construction of the complex asssociative algebras
is given , and some of the problems with the presentation of the classification in @xcite are discussed . since peirce
, there are other classifications of 3 , 4 and 5 dimensional associative algebras , like of b.g .
scorza , 1938 @xcite , p. gabriel @xcite and g. mazzola @xcite .
they all use different from ours techniques , and we rather stayed at the original peirce s classification .
the main issue , which we found to be problematic when we compared our classification with benjamin peirce s , is that he analyzes only those algebras which are called by him _ pure algebras _ , but does not give a definition of what he means by this concept .
charles peirce , in footnotes added in publication , partially explains what is meant by a non - pure algebra ; more clarification is provided in @xcite .
the fact remains that peirce did not classify the non - pure algebras , although we note that such a classification would not be difficult to achieve , so peirce s classification of the pure algebras does provide the key insight into the classification of complex associative algebras .
the main technique introduced by benjamin peirce is his proof that every complex algebra either contains an idempotent element , defined as an element @xmath2 such that @xmath3 , or every element is nilpotent , that is @xmath4 , for some @xmath5 . from this observation alone , he obtains a classification of complex algebras up through dimension 5 , so it is a quite powerful tool .
our approach to the classification problem is completely different from the existing approaches .
we use versal deformations to analyze the moduli space , and decompose it into strata , determined by a decomposition in terms of smooth and jump deformations .
this is a new approach to the study of moduli spaces of low dimensional algebras , which was first introduced in @xcite .
let @xmath6 be a vector space over a field .
then an associative algebra structure over is given by a -linear map @xmath7 , in other words , an element of @xmath8 .
if @xmath9 is the space of @xmath10-cochains , then the space of cochains @xmath11 can be expressed as a direct product @xmath12 . in @xcite ,
a bracket operation , the _ gerstenhaber bracket _ on @xmath13 was introduced , which equips the space of cochains with the structure of a -graded lie algebra , where an element of @xmath14 is odd when @xmath10 is even and even otherwise .
moreover @xmath15 is an element of @xmath16 , thus odd , and @xmath15 is an associative algebra structure precisely when @xmath17=0 $ ] .
jim stasheff in @xcite realized that one could identify @xmath13 with the coderivations of the tensor coalgebra @xmath18 in a natural manner , and that the bracket of cochains corresponds to the usual bracket of coderivations .
an odd coderivation @xmath15 satisfying @xmath17=0 $ ] is called a codifferential . thus associative algebra structures are simply codifferentials @xmath19 .
two associative algebras , given by the codifferentials @xmath15 and @xmath20 , are isomorphic when the codifferentials are equivalent under the following action of @xmath21 on @xmath13 given as follows .
any linear map @xmath22 induces a map @xmath23 by @xmath24 , which is bijective precisely when @xmath25 is invertible .
then @xmath25 acts on @xmath26 by @xmath27 .
if @xmath15 and @xmath20 are the two multiplication structures , then this definition of equivalence is the same as the usual notion of isomorphism , that is @xmath28 the set of equivalence classes of codifferentials in @xmath16 is called the moduli space of associative algebra structures on @xmath6 . in @xcite ,
the notion of _ hochschild cohomology _ of an associative algebra was defined .
it is given by a map @xmath29 which maps @xmath14 to @xmath30 .
gerstenhaber expressed this _
hochschild coboundary operator _
@xmath31 in terms of the gerstenhaber bracket , by @xmath32 $ ] . from this point of view
, the fact that @xmath33 follows immediately from the codifferential property @xmath34=0 $ ] . in a series of articles @xcite , gerstenhaber went on to use hochschild cohomology to study the deformation theory of associative algebras .
a more general definition of deformation was given in @xcite , in terms of a base given by a local , augmented , commutative algebra @xmath35 .
a local algebra is an algebra with a unique maximal ideal @xmath36 , and an augmentation is a morphism @xmath37 .
if @xmath35 is a local algebra with an augmentation , then the augmentation is unique , and is given by the map @xmath38 .
we shall call the algebra @xmath35 infinitesimal if @xmath39=0 , and complete if @xmath40 .
if @xmath6 is a -vector space , then @xmath41 is a free @xmath35-module , the tensor coalgebra @xmath42 of @xmath43 over @xmath35 can be identified with @xmath44 , and the space of cochains @xmath45 can be identified with @xmath46 .
( when @xmath35 is a complete local algebra , one uses @xmath47 and @xmath48 instead . ) note that the augmentation induces the structure of an @xmath35-module on any -vector space .
a deformation of an associative algebra structure on @xmath6 with base @xmath35 is a codifferential @xmath49 on @xmath35-module @xmath50 , satisfying the following property .
the map @xmath51 , given by @xmath52 must satisfy @xmath53 .
an automorphism of @xmath6 over @xmath35 is an @xmath35-linear map @xmath54 , such that @xmath55 is the identity map on @xmath6 . if @xmath56 for some automorphism of @xmath6 over @xmath35 , then the deformation @xmath57 is said to be equivalent to the deformation @xmath49 .
if @xmath35 is infinitesimal , then the deformation is called an infinitesimal deformation , and if @xmath35 is complete , the deformation is called a formal deformation .
this latter terminology comes from the fact that in the classical example of a 1-parameter formal deformation , the deformation is given as a formal power series in a parameter .
the classical example of an infinitesimal 1-parameter deformation can be described in terms of the infinitesimal base @xmath58/(t^2)$ ] , and is given by a codifferential of the form @xmath59 where @xmath15 is the codifferential giving the original multiplication .
the property that @xmath60 is a codifferential is simply @xmath61 . where @xmath31 is the hochschild coboundary operator . in terms of brackets of cochains ,
this property is @xmath62=0 $ ] .
the classical example of a formal 1-parameter deformation is one of the form @xmath63 with base @xmath64 $ ] .
the conditions which must be satisfied for a formal deformation can be expressed in terms of brackets of the cochains @xmath65 , and are @xmath66=0 $ ] . for @xmath67 , where @xmath68 .
if @xmath49 is a deformation with base @xmath35 , and @xmath69 is a morphism of @xmath70-algebras , then there is an induced deformation @xmath71 with base @xmath72 , given by @xmath73 .
this allows one to define the notion of a _ universal infinitesimal deformation _ @xmath49 , with infinitesimal base @xmath35 as follows .
the deformation @xmath49 is universal if , given any infinitesimal deformation @xmath71 with infinitesimal base @xmath72 , there is a unique morphism @xmath69 such that @xmath74 is _ infinitesimally equivalent _ to @xmath71 , in other words , equivalent as infinitesimal deformations of @xmath15 . in @xcite
, it was proven that there is a universal infinitesimal deformation for finite dimensional lie algebras , over a certain universal infinitesimal base .
the proof for associative algebras is essentially the same .
in general , there is no universal formal deformation , but there is a weaker notion of a versal deformation . in @xcite
a general framework was established for proving the existence of a versal deformation , which was applied to lie algebras in @xcite .
a construction of a versal deformation for lie algebras was given in @xcite , which carries over without any difficulties for associative algebras .
a generalization of this construction to the case of infinity algebras appeared in @xcite .
a versal deformation of an associative algebra given by the codifferential @xmath15 is a formal deformation @xmath49 of @xmath15 with base @xmath35 , such that if @xmath71 is a formal deformation of @xmath15 with base @xmath72 , then there is a morphism @xmath69 such that @xmath74 is _ formally equivalent _ to @xmath71 , where formal equivalence means equivalent as formal deformations with base @xmath72 . because the morphism @xmath75 is not in general , unique ,
a versal deformation is not universal .
the cohomology of a codifferential @xmath15 is @xmath76 , where @xmath31 is the hochschild coboundary operator determined by @xmath15 . since @xmath77
, we have a natural stratification of the cohomology by spaces @xmath78 note that in deformation theory , we are only interested in calculating cohomology with coefficients in @xmath6 , so our notation refers to this notion .
let @xmath79 be a prebasis of @xmath80 .
by prebasis , we mean a set of preimages of a basis of @xmath81 , lying in @xmath82 , also called a set of representative cocycles .
the universal infinitesimal deformation @xmath83 of a codifferential @xmath15 is of the form @xmath84 ( we are using the summation convention on @xmath85 ) , and @xmath86 are a set of infinitesimal parameters , meaning that @xmath87 in the infinitesimal base @xmath88/(t_jt_k)$ ] .
since the @xmath79 are cocycles , it follows that @xmath89=0 $ ] .
the infinitesimal deformation is also called the first order deformation . to construct the second order deformation ,
consider the bracket @xmath89 $ ] , but this time , without assuming any property of the parameters @xmath86 .
in general , there will be second order terms in this bracket .
however , it is easy to see that @xmath89 $ ] is a 3-cocycle , so if @xmath90 is a basis of @xmath91 and @xmath92 is a basis of the coboundaries @xmath93 , then @xmath94=a^{ij}_kt_it_j\alpha^k+b^{ij}_kt_it_j\beta^k.\ ] ] now the @xmath95 for some 2-cochains @xmath96 , and if we consider the second order deformation @xmath97 , it is easy to see that in the bracket @xmath98 $ ] , the second order terms involving @xmath99 drop out , and third order terms are added to the bracket .
the second order terms involving the @xmath100 do not drop out .
therefore , we assume that the polynomials @xmath101 must be equal to zero , up to third order , so for a base of the second order deformation is given by @xmath102/(r^k,\m^2)$ ] , where @xmath103 is the maximal ideal in @xmath102 $ ] . continuing in this fashion
, we eventually arrive at an expression for the versal deformation @xmath104 where the @xmath105 are formal power series of order at least two in the parameters @xmath86 , and the @xmath106 are a prebasis of the 3-coboundaries , , a preimage in @xmath82 of a basis of @xmath107 . by order ,
we mean the degree of the smallest term in the formal power series .
we also obtain a series of relations @xmath108 , which are also given by formal power series of order at least two , one relation for each basis element in @xmath91 .
thus the base of the formal deformations is @xmath102/(r_k)$ ] .
note that the number of parameters @xmath105 is equal to the dimension of the 3-coboundaries , while the number of parameters @xmath86 is equal to the dimension of @xmath81 .
this process of constructing the versal deformation order by order is quite tedious , although it often terminates rather quickly .
however , the process may not terminate at all , so we have been using the following method of computing the versal deformation .
we simply write the formula for the versal deformation above , with undetermined coefficients @xmath105 .
consider the bracket @xmath89 $ ] , and project this to the space of 3-coboundaries .
if the deformation is versal , this projection should be equal to zero .
if @xmath109 , then we obtain @xmath110 quadratic polynomials in the @xmath110 variables @xmath105 , in terms of the parameters @xmath86 . in good cases
, there should be a unique solution for the @xmath105 as functions of the parameters @xmath86 .
next , the projection of @xmath89 $ ] onto @xmath91 , will give @xmath111 polynomials in the variables @xmath86 and @xmath105 , where @xmath112 . substituting the solutions for the @xmath105 into these polynomials
give the @xmath111 relations on the base of the versal deformation .
this gives the versal deformation and the relations on the base .
note that the construction of the versal deformation depends on the choice of prebases for @xmath81 , @xmath91 , @xmath107 and @xmath113 as well as a choice of basis of @xmath107 .
the relations on the base and the form of the versal deformation depend on these choices , so a lucky choice can greatly simplify the construction .
now , how can we use the versal deformation to construct actual deformations , which are convergent formal power series ?
if one is lucky , the relations will turn out to be rational functions of the parameters @xmath86 , and it will be possible to solve these relations to yield simple solutions .
every such solution will determine a deformation of @xmath15 .
some of the solutions are bounded away from the origin . in this case , we say that the solution is not local , because it is not defined for small values of the parameters . local solutions to the relations on the base determine how the codifferential deforms to other codifferentials .
our method for constructing the moduli space as a geometric object is based on the idea that codifferentials which can be obtained by deformations with small parameters are `` close '' to each other . from the small deformations
, we can construct 1-parameter families or even multi - parameter families , which are defined for small values of the parameter , except possibly when the parameters vanish .
let us consider the 1-parameter case , since the multi - parameter case can be reduced to studying 1-parameter families .
if @xmath60 is a one parameter family of deformations , then two things can occur .
first , it may happen that for every small value of @xmath114 except zero @xmath60 is equivalent to a certain codifferential @xmath20 .
then we say that @xmath60 is a jump deformation from @xmath15 to @xmath20 .
it will never occur that @xmath20 is equivalent to @xmath15 , so there are no jump deformations from a codifferential to itself .
otherwise , the codifferentials @xmath60 will all be nonequivalent if @xmath114 is small enough . in this case , we say that @xmath60 is a smooth deformation . in @xcite
, it was proved for lie algebras that if one has three codifferentials @xmath15 , @xmath20 and @xmath115 , and there are jump deformations from @xmath15 to @xmath20 and from @xmath20 to @xmath115 , then there is a jump deformation from @xmath15 to @xmath115 .
similarly , if there is a jump deformation from @xmath15 to @xmath20 , and a family of smooth deformations @xmath116 , then there is a family @xmath60 of smooth deformations of @xmath15 , such that every deformation in the image of @xmath116 lies in the image of @xmath60 , for sufficiently small values of @xmath114 . in this case
, we say that the smooth deformation of @xmath15 factors through the jump deformation to @xmath20 . in the examples of complex moduli spaces of lie algebra which we have studied , it turns out that there is a natural stratification of the moduli space of @xmath0-dimensional lie algebras by orbifolds , where the codifferentials on a given strata are all connected by smooth deformations , which determine the local neighborhood structure .
the strata are connected by jump deformations , in the sense that any smooth deformation from a codifferential on one strata to another strata factors through a jump deformation .
moreover , all of the strata are given by projective orbifolds .
in fact , in all the complex examples we have studied , the orbifolds either are single points , @xmath117 , @xmath118 or @xmath119 . for higher dimensional complex lie algebras ,
we know that there are strata of the form @xmath120 in the moduli space of lie algebras of dimension @xmath121 .
we do nt have any proof , but we conjecture that this pattern holds in general . in other words , we believe the following conjecture . the moduli space of lie or associative algebras of a fixed finite dimension @xmath0 are stratified by projective orbifolds , with jump deformations and smooth deformations factoring through jump deformations providing the only deformations between the strata .
extensions of algebraic structures have been studied and classified by many people , for example @xcite . in @xcite
, we gave a treatment of this old problem that is useful in constructing equivalence classes of extensions , because it uses cohomology as the primary tool in the construction , so is well adapted to our computational methods .
we describe our approach to the extension problem below .
the notion of an extension of an associative algebra @xmath122 by an ideal @xmath110 is given by the exact sequence of algebras @xmath123 where @xmath124 .
let @xmath125 be the multiplication structure on @xmath110 and @xmath126 the multiplication structure on @xmath122
. then the multiplication structure @xmath15 on @xmath6 is given by two additional structures : the `` module structure '' @xmath127 , and the `` cocycle '' @xmath128 .
we have @xmath129 , and the fact that @xmath15 is an associative algebra structure is given by the codifferential property @xmath17=0 $ ] .
this fact is equivalent to the three conditions @xmath130=0\\ & [ \delta,\lambda]+\tfrac12[\lambda,\lambda]+[\mu,\psi]=0\\ & [ \delta+\lambda,\psi]=0.\end{aligned}\ ] ] if we define @xmath131 $ ] , then @xmath132 , so the first condition says that @xmath133 is a @xmath134-cocycle . under a natural notion of equivalence of extensions ,
cohomologous cocycles determine equivalent extensions .
this notion of equivalence is such that if two extensions are equivalent , then the codifferentials on the space @xmath6 are equivalent , so equivalent extensions determine the same element in the moduli space of codifferentials on @xmath6 . denoting the cohomology on the space of cochains @xmath11 of @xmath6 by @xmath135 , we note that the bracket on @xmath13 descends to a bracket on @xmath135 . denoting the image of a @xmath134-cocycle @xmath136 in @xmath135 by @xmath137
, we note that @xmath138 and @xmath139 are well defined .
we can define an operator @xmath140 on @xmath135 by @xmath141 $ ] . by the second condition , @xmath142 ,
denote the cohomology class of a @xmath140-cocycle @xmath137 in the cohomology @xmath143 induced by @xmath140 by @xmath144 $ ] .
we also note that @xmath134 commutes with @xmath140 on @xmath13 , which means that @xmath145 , so induces a cohomology on this subcomplex , which we denote by @xmath146 .
let @xmath147 be the subgroup of @xmath148 , consisting of those linear automorphisms @xmath25 of @xmath6 such that @xmath149 , @xmath150 , @xmath151 and @xmath152
. then @xmath153 also descends to an action on @xmath154 , so @xmath155 is well defined .
finally , let @xmath156 be the subgroup linear transformations of the form @xmath157 , where @xmath158 , and @xmath159 , such that @xmath160 $ ] .
these are the automorphisms of @xmath6 which have the property that the @xmath126 , @xmath125 and @xmath133 parts of @xmath15 are preserved under the action of @xmath161 , and are a subgroup of the group @xmath162 of automorphisms @xmath163 which preserve only @xmath126 and @xmath125 . in @xcite , the following classification of extensions
was proved .
the equivalence classes of extensions of a codifferential @xmath126 on @xmath122 by a codifferential @xmath125 on @xmath110 under the action of the group @xmath162 1 .
isomorphism classes of @xmath134-cohomology classes @xmath164 which satisfy the mc - equation @xmath165}=0\in h^{1,2}_\mu(\ker(d_{\dl}))\ ] ] under the action of the group @xmath147 on @xmath166 .
isomorphism classes of @xmath167-cohomology classes @xmath168\in h_{\mu,\dl}^{0,2}$ ] under the action of the group @xmath169 .
here @xmath170 is the subspace of @xmath154 given by the image of elements @xmath171 and similarly , @xmath172 is the subspace given by the image of @xmath173 in @xmath143 .
the cochain @xmath174 in the theorem arises in the following manner .
given a fixed @xmath133 , if @xmath175 is one solution to the conditions for an extension , then the set of all such solutions is of the form @xmath176 .
the cohomology class @xmath168 $ ] is well defined , and the equivalence class of the extension depends only on this cohomology class , with isomorphism classes giving equivalent extensions as stated in the theorem .
there is a nice prescription that one can follow in applying the theorem to construct extensions , with a minimal amount of duplication .
note that two extensions can be equivalent as associative algebras on @xmath6 without determining equivalent extensions , so some duplication can still occur in following the prescription . 1
. solve @xmath177 .
for a fixed @xmath125 , this gives linear constraints on @xmath133 .
2 . solve @xmath178 for @xmath179 .
simplify @xmath133 by removing coboundary terms .
solve the second condition on an extension .
this puts linear constraints on @xmath175 and quadratic constraints on @xmath133 .
4 . determine the group @xmath147 and apply generic elements to @xmath133 to simplify the choice of a representative .
now fix @xmath133 and determine @xmath156 .
solve the equation @xmath180=0 $ ] , which gives linear constraints on @xmath175 .
fix some @xmath175 giving an extension .
solve @xmath181=[\dl,\tau]=0 $ ] for @xmath174 .
. determine the group @xmath156 and apply it to @xmath174 to determine the equivalence classes of @xmath174 . to use this methodology to construct the moduli space ,
one simply begins with the first choice of @xmath126 and @xmath125 , determines equivalence classes of extensions , then further checks to see which ones are independent of the others , and then proceeds with additional choices for @xmath126 and @xmath125 .
all 3-dimensional complex associative algebras can be constructed as extensions of a 1-dimensional algebra by a 2-dimensional ideal .
this follows from the classification of simple associative algebras .
there are no simple 3-dimensional complex associative algebras , and thus every 3-dimensional associative algebra must contain a 2-dimensional ideal . since there are 7 nonequivalent 2-dimensional associative algebras and 2 nonequivalent 1-dimensional associative algebras , there are 14 cases which need to be checked .
the procedure is straightforward , and we were able to check them all easily .
in order to describe the types of associative algebras on a vector space @xmath122 , we introduce the following notation .
let @xmath182 be a basis of @xmath122 .
define @xmath183 by @xmath184 then any codifferential can be expressed in the form @xmath185 , where @xmath186 are called the structure constants of the algebra .
it is also convenient to represent @xmath15 in terms of a matrix .
we consider the basis of @xmath187 given by @xmath188 .
then @xmath15 is given in terms of this input basis and the standard output basis by an @xmath189 matrix @xmath190 .
it is often easier to see what is going on when studying the deformations of a codifferential to look at the matrix of the deformed algebra .
there are only two 1-dimensional complex algebras , the trivial algebra structure , and the structure given by the codifferential @xmath191 , which is the ordinary multiplication on @xmath192 .
we will denote the complex numbers equipped with the first structure by @xmath193 , and with the second structure by @xmath194 .
the types of 2-dimensional complex associative algebras are given in table [ table 1 ] below .
[ table 1 ] @xmath195 the classification of the algebras can already be determined from @xcite , which explicitly gives @xmath196 , @xmath197 and @xmath198 , and implicitly @xmath199 . the algebras @xmath200 and @xmath201 are decomposable as direct sums .
we will denote @xmath202 , equipped with the algebra structure given by the codifferential @xmath203 as @xmath204 .
thus @xmath205 , and @xmath206 gives a representation of the structures @xmath200 and @xmath201 as direct sums .
of these algebras , only @xmath196 and @xmath199 are not commutative . in terms of peirce s classification by nilpotent and idempotent elements ,
types @xmath196 , @xmath199 and @xmath197 contain nonzero idempotents , while @xmath197 is a nilpotent algebra .
the algebras @xmath201 and @xmath207/(x^2)$ ] are the only unital algebras .
a complete list of the 22 distinct types of 3-dimensional complex associative algebras is given in table [ table 2 ] .
@xmath208 there is a family of algebras labeled @xmath209 , where the parameter @xmath210 in @xmath211 is a projective coordinate , representing the fact that any nonzero multiple of the parameter gives an equivalent deformation .
actually , @xmath212 , so that this family of differentials is parameterized by @xmath118 , where the action of the permutation group @xmath213 on @xmath117 is by interchanging coordinates .
the codifferential @xmath214 is expressed as an extension of a @xmath215-dimensional algebra by a @xmath216-dimensional algebra .
however , it is equivalent to the codifferential @xmath217 which is an extension of a @xmath216-dimensional algebra by a @xmath215-dimensional algebra .
the reason it is given in the form in which it appears in the table is that the cohomology was much simpler to compute in this form .
however , the fact that it is equivalent to the codifferential above implies that every 3-dimensional associative algebra has a 2-dimensional ideal .
a partial classification of 3-dimensional associative algebras appears already in @xcite .
our method of obtaining these codifferentials differs from peirce s in that we constructed them by studying extensions of @xmath215-dimensional associative algebras by @xmath216-dimensional algebras ( and vice - versa ) . by wedderburn s theorem ,
there are no simple 3-dimensional complex associative algebras , so all three dimensional complex associative algebras arise as extensions .
the algebras @xmath218 , as well as @xmath219 , @xmath220 and @xmath221 are direct sums , while the algebras @xmath222 , @xmath223 , @xmath224 , @xmath225 and @xmath226 are examples of what peirce calls not pure algebras .
the algebras @xmath227 , @xmath214 , @xmath228 and @xmath211 all correspond to algebras on peirce s list . in a footnote in @xcite , added by peirce s son charles peirce ,
there is an explanation that a not pure algebra is one in which there is a basis @xmath2 , @xmath229 and @xmath230 such that @xmath2 and @xmath229 span a subalgebra , @xmath2 and @xmath230 span another , and @xmath231 . if one assumes that products of @xmath2 and @xmath229 are multiples of @xmath229 and that products of @xmath2 and @xmath230 are multiples of @xmath230 , then one obtains automatically the structure of an associative algebra , whenever the two subalgebras are associative
of course , the algebras which are direct sums are also not pure in this sense .
peirce does not give a classification of the algebras which are not pure , confining himself to classifying the pure algebras .
however , it would probably not be a difficult task to classify the not pure algebras using the methods in peirce s paper .
thus , our analysis concurs with peirce s classification . to compute the nonequivalent codifferentials
, we used the methods outlined in @xcite , which outlines a method of computing nonequivalent extensions using cohomological methods .
the codifferential @xmath191 represents the multiplication structure @xmath232 . since @xmath233 and @xmath234
, one might expect that the versal deformation would be quite complicated . however , it turns out that the infinitesimal deformation coincides with the versal deformation .
the matrix of the versal deformation is given by @xmath235.\ ] ] there are 16 relations on the base of the versal deformation , which we omit , for sake of brevity .
although the relations are numerous , their solution is not so complicated . using maple , we found the following seven solutions to the relations .
@xmath236 in order to determine which codifferential a deformation is equivalent to , one needs to substitute a solution to the relations into the general form for the versal deformation .
thus we studied 7 versions of the versal deformation , corresponding to the 7 solutions above .
we omit giving details about which codifferentials each of the 7 turns out to be equivalent to .
what is important is which codifferentials arise as deformations for small values of the parameters .
note each of the solutions above remains valid if we substitute all of the parameters with zero . such a solution will be called _
local_. only the local solutions are relevant to the description of how the moduli space is glued together .
we found that @xmath200 deforms to types @xmath196 , @xmath198 , @xmath237 , @xmath238 , @xmath239 , and @xmath220 .
the codifferential @xmath196 represents the multiplication structure @xmath240 . since @xmath241 ,
the versal deformation is not so difficult to compute , and it turns out that the infinitesimal deformation is versal in this case as well ; moreover , the relations on the base vanish , so that the coderivation @xmath242 , represented by the matrix @xmath243\ ] ] is indeed a codifferential for all values of the parameters .
we found that @xmath196 deforms to @xmath198 , @xmath239 and @xmath220 .
the codifferential represents the multiplication structure @xmath244 .
we have @xmath241 , and the versal deformation is given by the infinitesimal deformation .
the matrix of @xmath242 is given by @xmath245.\ ] ] there is one nontrivial relation @xmath246 , which has the two solutions @xmath247 and @xmath248 , both of which are local .
we found that @xmath199 deforms to @xmath238 and @xmath223 .
the codifferential @xmath201 represents the multiplication structure @xmath249 .
the infinitesimal deformation is versal , and is given by the matrix @xmath250.\ ] ] there is one nontrivial relation @xmath251 .
we found that @xmath201 deforms to types @xmath237 and @xmath223 .
note that @xmath199 and @xmath201 are mirror images of each other , so they should have the same type of pattern to their deformations .
the codifferential @xmath197 represents the multiplication structure @xmath252 .
the infinitesimal deformation , given by the matrix @xmath253,\ ] ] is versal and there are no relations .
we found that @xmath197 deforms to types @xmath198 , @xmath239 and @xmath220 .
the codifferential @xmath198 represents the multiplication structure @xmath254 .
the matrix of the infinitesimal deformation , which is versal , is given by @xmath255.\ ] ] there are no relations on the base .
we found that @xmath198 deforms only to type @xmath220 .
the codifferential @xmath256 does not decompose as a direct sum . since @xmath257 , there are no deformations .
note that the cohomology vanishes in even degrees , and has dimension @xmath216 in odd degrees .
therefore , there are no odd elements
. the codifferential @xmath258 does not decompose as a direct sum . since @xmath257 , there are no deformations . in this case , only @xmath259 does not vanish .
the codifferential @xmath260 does not decompose as a direct sum .
its infinitesimal deformation , which is versal , is given by the matrix @xmath261.\ ] ] we found that @xmath260 deforms only to @xmath237 .
the codifferential @xmath237 represents the multiplication structure @xmath262 . since @xmath257
, it does not deform .
the codifferential @xmath238 represents the multiplication structure @xmath263 . since @xmath257 it does not deform .
the codifferential @xmath239 does not decompose as a direct sum .
its infinitesimal deformation , given by the matrix @xmath264,\ ] ] is versal , and there are no relations .
we found that @xmath239 deforms only to @xmath220
. the codifferential @xmath223 does not decompose as a direct sum , and it has no deformations , since @xmath257 .
the codifferential @xmath220 decomposes as @xmath265 .
there are no deformations , since @xmath257 .
the codifferential @xmath221 represents the multiplication structure @xmath266 .
since the dimension of @xmath81 is 9 , the largest dimension of @xmath81 for any of the codifferentials , it is not surprising that the versal deformation is quite complex .
its infinitesimal deformation is not versal .
we shall omit giving the matrix of the versal deformation , but mention that it is given in terms of at most cubic polynomials in the parameters , so that the third order deformation is versal .
we also omit the 17 relations on the base of the versal deformation .
we note that with the help of maple , we decomposed the relations into 9 solutions , all of which are local .
we found that @xmath221 deforms to types @xmath200 , @xmath196 , @xmath199 , @xmath201 , @xmath197 , @xmath198 , @xmath260 , @xmath237 , @xmath238 , @xmath239 , @xmath223 , @xmath220 , @xmath227,@xmath224 , @xmath225 , @xmath214 , @xmath228 and @xmath209 for every value of the parameter @xmath210 except @xmath267 .
in other words , it deforms to all of the codifferentials except for @xmath256 , @xmath258 , @xmath226 , and @xmath268 .
the codifferential @xmath227 is not a direct sum , and one of the algebra s given in peirce s article .
the infiniitesimal deformation is versal and has matrix @xmath269\ ] ] we found that @xmath227 deforms to @xmath196 , @xmath197 , @xmath198 , @xmath239 @xmath220 and @xmath214 .
the codifferential @xmath224 is not a direct sum .
we found that @xmath224 deforms only to type @xmath238 .
the codifferential @xmath225 is not a direct sum . since @xmath270 and @xmath271 , it is not surprising that there might be some difficulty computing the versal deformation , and indeed , we were unable to compute the relations on the base of the versal deformation explicitly
. nevertheless , we were able to determine all the local solutions to the relations and determine which codifferentials they were equivalent to .
we found that @xmath225 deforms to types @xmath239 , @xmath223 , @xmath220 and @xmath214 .
the codifferential @xmath214 is not a direct sum , and it appears in peirce s classification .
its infinitesimal deformation is versal , and is given by the matrix @xmath272.\ ] ] we found that @xmath214 deforms to types @xmath239 and @xmath220 .
the codifferential @xmath226 is not a direct sum .
since @xmath257 , it has no deformations .
the codifferential @xmath228 is not a direct sum , and it appears in peirce s classification .
the relations on the base of the versal deformation are @xmath273 note that these expressions are rational , not polynomial , in the @xmath114-parameters . as a consequence , calculation of the versal deformation order by order would not be successful . however , there are only two solutions to these equations , given by @xmath274 the first of these is not local .
the other solution give a very simple matrix for the versal deformation : @xmath275.\ ] ] thus , even though @xmath81 has dimension 3 , effectively it as is if the dimension was only 1 , because the relations force such a simplification in any true deformation .
we found that @xmath228 deforms to @xmath209 , except for the values @xmath276 , @xmath277 and @xmath267 .
the exceptions turn out to be important , because for these exceptional values of the parameter @xmath210 , the codifferential @xmath211 has extra deformations .
since @xmath228 does not have any extra deformations , it would be impossible for it to deform into these types .
this is not the reason we know it has no such deformations ; rather we were gratified to see that our computations of the deformations of @xmath228 did not lead to any contradictions .
the codifferential @xmath209 is not a direct sum , and appears as a family on peirce s list .
more precisely , peirce gives a 1-parameter family of associative algebra structures , parameterized by @xmath2 , which mostly coincides with our family .
the case @xmath278 , which peirce points out as being special , coincides with our codifferential @xmath214 .
also @xmath279 corresponds to the codifferential @xmath214 , and more generally , the multiplication determined by any parameter @xmath2 is equivalent to that determined by @xmath280 , although this fact was not mentioned in @xcite .
the coderivations @xmath281 and @xmath268 are not equivalent to members of peirce s family , but all the rest are . as was mentioned in the beginning of the section , the family @xmath209 is parameterized by @xmath282 .
note that peirce s family is parameterized by @xmath283 , where @xmath213 acts by taking @xmath2 to @xmath280 .
topologically , the resulting quotient is still @xmath192 , and if we remove the point corresponding to @xmath279 , we obtain @xmath117 minus two points . since the orbifold @xmath118 which parameterizes @xmath211 is topologically @xmath117 , removing the points @xmath276 and @xmath267 gives the same topological type as peirce s family .
generically , the dimension of @xmath91 is 3 , but for the special cases @xmath276 , @xmath277 , and @xmath267 the cohomology has dimension 5 , so these cases require separate treatment .
let us describe the generic case first .
the matrix of the versal deformation is complicated , so we omit it .
the relations on the base of the versal deformation are @xmath284 with two solutions : @xmath285 the second solution is not local , unless @xmath286 , which is the special case @xmath287 to be considered later .
thus , we can restrict our attention to the first solution , in which case the versal deformation is simply @xmath288 .
this is the first case we have studied in which a deformation occurs which is not a jump deformation .
this smooth deformation is simply a deformation along the family .
thus generically , a codifferential @xmath209 only deforms along the family @xmath211 , and it deforms smoothly along the family . for this special case , the base of the versal deformation is given by 8 relations , two of which are trivial .
solving the relations , one obtains 6 solutions , only 4 of which turn out to be local .
we found that @xmath281 deforms to types @xmath199 , @xmath201 , @xmath260 , @xmath237 , @xmath238 , @xmath223 and @xmath224 , as well as smoothly deforming along the family .
for this special case , the base of the versal deformation is given by 7 relations , which yield 3 solutions , 2 of which are local .
we found that @xmath289 deforms to types @xmath196 , @xmath197 , @xmath198 , @xmath239 , @xmath220 , @xmath227 and @xmath214 .
once you know that it deforms to type @xmath227 , this determines all the rest , because @xmath227 deforms to all the other ones .
we also found that @xmath289 deforms along the family @xmath211 .
for this special case , the base of the versal deformations is given by 8 relations , which are very complicated , but still only give 3 solutions , 2 of which are local . as well as deforming along the family , we find that @xmath287 also deforms to type @xmath256 .
let us define the level of an algebra as follows .
a codifferential is on level 0 if it has no deformations .
a codifferential is on level @xmath290 if some element of its family has a deformation to an element on level @xmath10 , and other than deformations along its own family , all deformations are to elements of level @xmath10 or below . using this definition of level , there are 7 levels for the elements in our moduli space
. on level 0 , there are 7 elements : @xmath256 , @xmath258 , @xmath237 , @xmath238 , @xmath223 , @xmath220 and @xmath226 . of these , @xmath258 and @xmath226 do nt even have any jump deformations to them from higher level objects , so they are truly isolated . on level 1 ,
there are 6 elements : @xmath199 , @xmath201 , @xmath198 , @xmath260 , @xmath239 and @xmath224 .
we have @xmath199 deforming to both @xmath238 and @xmath223 , @xmath201 to both @xmath237 and @xmath223 , @xmath198 deforms to @xmath220 only , @xmath260 deforms only to @xmath237 , @xmath239 to @xmath220 only , and @xmath224 deforms only to @xmath238 . on level 2 , there are 3 elements : @xmath196 , @xmath197 and @xmath214 .
both @xmath196 and @xmath197 deform to @xmath198 and @xmath239 on level 1 , and @xmath214 deforms to @xmath239 on level 1 .
on level 3 , there are 3 elements : @xmath200 , @xmath227 and @xmath225 . here
the picture is more complex , because the jumps to lower levels do not always factor through a jump to level 2 .
we have @xmath200 deforming to @xmath196 on level 2 , as well as directly to @xmath237 and @xmath238 on level 0 .
the behaviour of @xmath227 is determined by jumps to @xmath196 , @xmath197 and @xmath214 on level 2 .
finally , @xmath225 deforms to @xmath214 on level 2 , as well as to @xmath223 on level 0 . on level 4 , there is only the family @xmath209 . while most elements deform only along the family , the special element @xmath289 deforms to @xmath227 on level 3 , which puts the family on level 4 .
in addition , @xmath281 deforms to @xmath199 , @xmath201 , @xmath260 and @xmath224 on level 1 , while @xmath268 has a jump to @xmath256 on level 0 . on level 5
, there is only the element @xmath228 , which has jump deformations to most elements of the family @xmath209 on level 4 .
finally , on level 6 , there is only @xmath221 , which has a jump to @xmath228 on level 5 , as well as jumps to almost all the other codifferentials . the moduli space of three dimensional complex associative algebras is illustrated in figure [ figure1 ] below .
the authors have been studying moduli spaces of low dimensional lie algebras , and have noticed that the description of the moduli space can be given in terms of a stratification by complex projective orbifolds , with families given by very special orbifolds of the type @xmath120 .
there are a lot of similarities between the picture for three dimensional associative algebras and the lie algebra case .
there is one family , parameterized by @xmath118 , and some special points , for both the three dimensional lie and the three dimensional associative algebra cases .
however , there are only 6 special points in the lie case , and 21 special points in the associative algebra case . in the lie case , there are no elements which have the property that they neither deform , nor does any element deform to them , but in the associative case , there are two such elements .
so there are similarities and differences in the pictures of lie and associative algebras .
the authors would like to thank jim stasheff , who read this manuscript and made many helpful remarks .
in addition , the authors thank mitch phillipson , an undergraduate student , for creating the illustration in figure 1 .
we also thank the referee for providing useful historical references .
to3em , _ an example of formal deformations of lie algebras _ , nato conference on deformation theory of algebras and applications ( dordrecht ) , kluwer , 1988 , proceedings of a nato conference held in il ciocco , italy , 1986 , pp . | in this paper , we give a classification of the 3-dimensional associative algebras over the complex numbers , including a construction of the moduli space , using versal deformations to determine how the space is glued together . |
the determination of parameters like critical exponents from finite size scaling @xcite is plagued by the appearance of corrections to scaling .
consider , e.g. , an ising model on a cubic lattice of linear extension @xmath8 .
the binder cumulant q= behaves at the critical point like [ qfss ] q(l)= q^ * + q l^- + @xmath9 denotes the magnetization per spin .
@xmath10 is universal .
@xmath11 denotes the leading correction to scaling exponent .
many more terms appear in eq .
( [ qfss ] ) , governed by subleading exponents and combinations of them .
if they have non - negligible coefficients in front of them , they can hamper or even make impossible a reliable determination of the infinite @xmath8 behaviour . in the language of the renormalization group the presence of strong corrections to scaling
is related to the fact that one is far away from the renormalization group fixed point . in this article
, we summarize the results of our attempt to improve the scaling properties of the 3d ising model by suitably tuning the parameters in a two - coupling spin-1 model that has already been studied by blte et al . @xcite
. we will be mainly interested in the determination of the exponents @xmath12 and @xmath13 that will be obtained from the scaling laws ( @xmath14 ) = a_i l^1/ ( 1 + b_i l^- + ) and [ chifit1 ] = c + l^2- ( 1 + d l^- + ) . here , @xmath15 denotes the cumulant introduced above , and @xmath16 is the ratio of partition functions @xmath17 , where the subscripts label periodic and antiperiodic boundary conditions , respectively .
antiperiodic boundary conditions are applied in one of the three lattice directions only .
the susceptibility @xmath18 is defined by = l^3 m^2 .
following @xcite , we consider a spin-1 ising model on a simple cubic lattice defined through s = - _ < i , j > s_i s_j + d _ i s_i^2 .
the @xmath19 take values @xmath0 , and the spin - spin interaction is a sum over all nearest neighbour pairs .
the boltzmann factor is given by @xmath20 .
blte et al .
chose a fixed @xmath21 , and varied @xmath22 to tune to criticality .
we followed a different procedure , to be described in some detail elsewhere . from prior simulations @xcite ,
estimates of the universal quantities @xmath23 were available , namely @xmath24 and @xmath25 .
for a number of lattice sizes @xmath26 ranging from 3 to 14 a sequence of couplings @xmath27 was determined such that @xmath28 within the statistical accuracy .
as @xmath8 increases the pair @xmath27 converges to a point where leading order corrections to scaling vanish . from our numerical results we extrapolate to @xmath29 .
we found that this point is approached on the line [ critline ] d -2.04 + 6.95 .
it turns out that eq .
( [ critline ] ) is an approximation to the critical line .
it is reassuring that the estimate @xmath30 , together with @xmath31 in @xcite is close to this line .
our next goal was to approximately identify a second one - dimensional submanyfold of the 2-coupling space , namely that where the leading correction to scaling vanishes .
this was possible by fitting the derivatives of the @xmath32 at @xmath33 with respect to the two coupling parameters to a certain linear equation .
details will be reported in @xcite .
our fit results suggested that the critical line should be approached by varying @xmath22 while adjusting @xmath34 according to [ goodline ] d = 3 ( -0.3832 ) + 0.6241 . a first estimate for @xmath35
was then obtained applying the well established method looking for the crossings of @xmath36 with @xmath37 .
we used lattices of size @xmath38 and obtained @xmath39 , with @xmath34 given by eq .
( [ goodline ] ) , i.e. @xmath40 .
this is significantly different from the @xmath31 used in ref .
monte carlo simulations were now performed at @xmath41 , fixing @xmath34 according to eq .
( [ goodline ] ) .
we simulated on cubic lattices with linear extension @xmath8 ranging from 4 to 56 .
state of the art cluster algorithms were employed .
two measurements were separated by three single cluster updates . for the computation of @xmath17 a variant of the boundary flip cluster algorithm @xcite was implemented .
the runs for the determination of the critical parameters described in the previous section took about 3 months of cpu on pentium 166-mmx pcs , while the final production runs consumed a total of 1 year .
the data relevant for the present article are summarized in tables [ zqx ] and [ deri ] .
table [ zqx ] gives the quantities @xmath42 , @xmath14 and @xmath43 .
in addition , we quote the number of measurements in the last column . in table [ deri ] we give the estimates of the partial and total derivatives of the @xmath42 with respect to @xmath22 .
note that eq .
( [ goodline ] ) implies @xmath44 the quantities in table [ deri ] are multiplied by a factor @xmath45 , which compensates to good precision for the leading divergent behaviour with increasing lattice size . by a reweighting technique we have access also to the same set of observables at a set of four neighbouring @xmath22 values , ranging from 0.383225 to 0.383265 in steps of 0.000010 .
having control over a neighbourhood of the simulation point is important , since we started without a high precision estimate of @xmath35 . [
cols=">,^,^,<,^",options="header " , ]
we have demonstrated that the spin-1 ising model with suitably chosen coupling constants has remarkably improved finite size scaling properties .
this allowed us to obtain high precision estimates for the critical indices @xmath12 and @xmath13 and two other universal quantities , the binder cumulant @xmath46 and the ratio of partition functions @xmath17 .
our results compare very well with a selection of estimates obtained by other authors , quoted in table [ exp1 ] .
many more estimates can be found in refs .
@xcite and @xcite .
our estimate for the cumulant , @xmath47 may be compared with the one given in @xcite , namely @xmath48 .
of course , one could also search for even further reduction of corrections to scaling by refining our procedure or choosing variants of the model , e.g. , the @xmath49-model or an ising model with more couplings .
last but not least , application of the ideas underlying the present analysis to other models seems promising . | we present results from the simulation of a two - coupling spin-1 model with states @xmath0 and nearest neighbour interaction . by a suitable choice of couplings we are able to drastically reduce the effects of corrections to scaling .
our estimates for the critical exponents are @xmath1 and @xmath2 . for the universal ratio @xmath3
we obtain @xmath4 . the universal ratio of partition functions with antiperiodic / periodic boundary conditions , respectively , is @xmath5 . 0.5 cm hub - ep-98/26 + ms
tpi986 1.5 cm 1.5 cm m. hasenbusch@xmath6 , k. pinn@xmath7 , and s. vinti@xmath7 + @xmath6 fachbereich physik , humboldt - universitt zu berlin + invalidenstr . 110 , d10099 berlin , germany + e mail : [email protected]_ @xmath7 institut fr theoretische physik i , universitt mnster + wilhelm klemm str . 9 , d48149 mnster , germany + e mail :
_ [email protected] , [email protected]_ 2.5 cm |
heavy quarks are regarded as effective probes of the strongly - interacting medium produced in ultra - relativistic heavy - ion collisions , since they are produced in the initial hard - scattering processes and they may subsequently interact with the medium itself . at the relativistic heavy ion collider ( rhic ) , a significant suppression of the so called ` non - photonic electrons ' , expected to be produced in the semi - leptonic decay of charm and beauty hadrons , has been measured in central au
au collisions at centre - of - mass energy @xmath8 per nucleon nucleon collision , indicating a substantial energy loss of heavy quarks in a strongly - interacting medium @xcite . in collisions at the large hadron collider ( lhc ) , the energy per nucleon nucleon collision will be of @xmath9 , about 30 times larger than at rhic , opening up a new era for the study of strongly - interacting matter at high energy density ( ` qcd medium ' )
. heavy - quark medium - induced energy loss will be one of the most captivating topics to be addressed in this novel energy domain @xcite . in hadron
hadron collisions at lhc energies , muons are predominantly produced in semi - leptonic decays of heavy - flavoured hadrons mostly beauty for muon @xmath10 .
thus , in heavy - ion collisions , the muon @xmath5 distribution is sensitive to b - quark energy loss effects . in absence of nuclear effects , the initial heavy - quark production yields are expected to scale from proton proton ( pp ) to nucleus
nucleus collisions in a given centrality class , according to the average number @xmath11 of inelastic nucleon nucleon collisions ( binary scaling ) . under this assumption ,
the in - medium energy loss of heavy quarks would induce a suppression of the high-@xmath5 muon yield with respect to the binary - scaled yield measured in pp collisions .
the suppression can be quantified as a reduction with respect to unity of the nuclear modification factor : @xmath12 however , initial - state effects , like for example nuclear shadowing of the parton distribution functions , could significantly reduce the initial production yields in nucleus
nucleus collisions , thus making more difficult to relate a reduction of @xmath0 to b - quark energy loss . at the lhc , muons from w and z decays can provide an intrinsic calibration for the muon nuclear modification factor and a test of the binary scaling assumption . due to the large amount of energy available at the lhc , w and z bosons will be produced with significant cross sections in the hard parton
parton scatterings , and their initial yields are expected to scale with @xmath11 .
as we will show , the muons from the decays @xmath13 and @xmath14 are predicted to dominate the muon @xmath7 distribution for @xmath15gev/@xmath16 .
three experiments , alice @xcite , atlas @xcite , and cms @xcite , will measure the production of muons in heavy - ion collisions , covering different acceptance regions .
the alice muon spectrometer covers the pseudo - rapidity range @xmath17 for @xmath18 ( @xmath19 ) .
atlas and cms can measure muons at central pseudo - rapidity , @xmath20 , with a larger cutoff , @xmath21@xmath22 . as we will quantify in section [ sec : results ] , the high-@xmath5 reach for the measurement of the inclusive muon production spectrum is expected to extend well into the region where muons from w and z decays become dominant over muons from beauty decays . in this work ,
we study the effect of heavy - quark energy loss on the transverse momentum distribution of muons in collisions at @xmath23 within the acceptance of the lhc experiments .
the weak gauge bosons contributions ( section [ sec : wz ] ) to the muon @xmath7 and rapidity distributions are obtained from the pythia @xcite event generator , taking into account the isospin content of the colliding nuclei and normalizing the cross section to the values predicted by the calculations in refs .
@xcite . the heavy quark contributions ( section [ sec : hq ] ) to the muon @xmath5 distribution are obtained from a nlo perturbative qcd(pqcd ) calculation ( mnr @xcite ) supplemented with the mass - dependent bdmps quenching weights for radiative energy loss @xcite , quark fragmentation _ la _ peterson @xcite and semi - muonic decay with the spectator model @xcite . in the calculation of heavy - quark energy loss the decreased medium density at large rapidities is taken into account by assuming the bdmps transport coefficient @xmath24 to scale as @xmath25 ( section [ sec : eloss ] ) . in section
[ sec : results ] we present and discuss the resulting muon @xmath5 distribution in the 10% most central collisions , the -to - pp nuclear modification factor @xmath26 and the central - to - peripheral nuclear modification factor @xmath27 , without and with the inclusion of heavy - quark energy loss .
the muon @xmath5 distributions in pp and collisions at lhc energies are calculated considering the semi - muonic decays of heavy - flavoured hadrons and the muonic decay of w and z bosons .
the procedure used to evaluate the production differential cross section per nucleon
nucleon collision is described in the following . at the lhc , the c.m.s .
energy is large enough to allow the production of massive particles such as the w and z bosons .
the @xmath7 and rapidity distributions of w / z and of their decay muons are obtained from the pythia event generator @xcite , that reproduces the measured @xmath5 distributions in @xmath30 collisions at the tevatron ( @xmath31 ) @xcite . since in hadron
hadron collisions , at leading order , w and z bosons are produced by quark
anti - quark annihilation , there could be a dependence of their production cross section on the isospin of the input channel . in nucleus
nucleus collisions this dependence can be mimicked by a weighted cocktail of proton proton ( pp ) , neutron neutron ( nn ) , proton neutron ( pn ) and neutron proton ( np ) collisions .
the cross section per nucleon
nucleon binary collision can be expressed as @xmath32 where @xmath33 and @xmath34 are the mass number and the atomic number of the colliding nuclei .
cteq4l @xcite parton distributions functions ( pdfs ) are used , and nuclear shadowing is accounted for via the eks98 parametrization @xcite .
the resulting @xmath5 distributions are normalized to the cross sections obtained from refs .
@xcite , that is a cross section per nucleon nucleon collision of @xmath35 ( @xmath36 ) nb for the w and @xmath37 ( @xmath38 ) nb for the z in ( pp ) collisions at @xmath39 tev , including the muonic branching ratios ( 10.6% for w and 3.4% for z @xcite ) . notice that the z production cross section is about ten times smaller than that of the w @xcite .
the uncertainty due to neglecting higher order corrections ( next - to - next - to - leading order ) was quantified as 12% in ref .
@xcite by varying the values of the factorization and renormalization scales .
the uncertainty due to the errors on the pdfs was quantified as about 10% in ref .
@xcite . within the pqcd collinear factorization framework
, the expression for the production cross section of heavy - flavoured hadrons in the collision of two hadrons @xmath33 and @xmath40 can be schematically written as : @xmath41 where @xmath42 and @xmath43 are the parton distribution functions , the differential probabilities for the partons @xmath44 and @xmath45 to carry momentum fractions @xmath46 and @xmath47 of their respective nucleons .
@xmath48 is the cross section of the partonic process @xmath49 .
the fragmentation function @xmath50 is the probability for the heavy quark q to fragment into a hadron @xmath51 with transverse momentum @xmath52 . to simplify the notation ,
we have dropped the @xmath53 dependence of @xmath48 and the renormalization / factorization scale dependences of @xmath54 , @xmath48 and @xmath55 ( the squares of scales are normally of the order of the momentum transfer @xmath56 of the hard scattering ) .
we use the nlo pqcd calculation implemented in the hvqmnr program @xcite to obtain the heavy - quark @xmath5@xmath57 double - differential cross sections , with the following parameters values : for charm , @xmath58 and factorization and renormalization scales @xmath59 , where @xmath60 ; for beauty , @xmath61 and @xmath62 .
cteq4 m @xcite parton distribution functions are used , and nuclear shadowing is taken into account with the eks98 parametrization @xcite . for the b quark ,
the perturbative uncertainty was quantified , by varying the scales , in about 30% for @xmath63 @xcite .
starting from the heavy - quark double - differential cross sections at nlo , we obtain the muon - level cross sections using the following monte carlo procedure .
we sample @xmath5 and @xmath57 of a c ( or b ) quark according to the shape of the nlo cross section and fragment it to a hadron using the peterson @xcite fragmentation function following the parameterization obtained in a recent analysis of @xmath64 data from lep @xcite . finally , we decay the hadron into a muon according to the spectator model @xcite . in the spectator model , the heavy quark in a meson
is considered to be independent of the light quark and is decayed as a free particle according to the @xmath65 weak interaction @xcite .
thus , we assume the momentum of the hadron to be entirely carried by the constituent heavy quark and we perform the heavy - quark three - body decay , @xmath66 or @xmath67 , to obtain the muon transverse momentum and rapidity .
the muon production cross sections per nucleon nucleon collision from charm ( beauty ) at @xmath23 that we obtain are 0.415 mb ( 20 @xmath68b ) in collisions and 0.637 mb ( 23 @xmath68b ) in pp collisions .
a charm ( beauty ) semi - muonic branching ratio of 9.6% ( 11.0% ) @xcite has been considered .
we do not include the muons from the cascade decay @xmath69 , because their yield is expected to become negligible with respect to that of the direct muons from b , for @xmath5 larger than a few @xmath6 @xcite .
the uncertainty on the beauty component , which dominates the heavy - flavour muon yields in the @xmath5 range relevant to our study , can be quantified as about 30% , on the basis of a recent analysis @xcite that considered the perturbative uncertainty ( variation of the factorization and renormalization scales ) , the uncertainty on the pdfs and the uncertainty on the fragmentation .
now we will compute the muon @xmath5 distribution taking into account the heavy - quark energy loss in the strongly - interacting medium that is expected to be formed in central collisions at lhc energies . for modelling the energy loss of heavy quarks by medium - induced gluon radiation , we used the quenching weights in the multiple soft scattering approximation , which were derived in ref .
@xcite in the framework of the bdmps formalism @xcite .
schematically , energy loss is introduced by modifying eq .
( [ eq : factorization ] ) to : @xmath70 where @xmath71 is the heavy - quark energy and @xmath72 is the radiated energy .
the quenching weight , represented by @xmath73 , is the probability for a heavy quark with mass @xmath74 and energy @xmath71 to lose an energy @xmath72 while propagating over a path length @xmath75 inside a medium with transport coefficient @xmath24 .
the latter is defined as @xmath76 , the average transverse momentum , @xmath77 , squared transferred from the medium to the parton per unit mean free path @xmath78 .
it is expected to be proportional to the volume density @xmath79 of gluons in the medium ( thus , to its energy density ) and to the typical momentum transfer per scattering .
we calculate the energy loss @xmath72 following the monte carlo approach introduced in ref .
@xcite for light quarks and gluons , and adapted for heavy quarks in refs .
we start by sampling the heavy - quark kinematics , @xmath5 and @xmath57 , according to the nlo double - differential cross section .
then , we sample the parton production point in the transverse plane @xmath80 according to the glauber - model @xcite density @xmath81 of binary collisions , and we sample the azimuthal parton propagation direction .
we calculate the path length @xmath75 and the value of @xmath24 , which is a mean value of the local time - averaged ( @xmath82 ) transport coefficient @xmath83 along the path of the parton .
we do not include the expansion of the medium in the longitudinal and transverse directions .
however , it has been shown in ref .
@xcite that , numerically , the effects of a time - dependent medium on parton energy loss can be accounted for by an equivalent static medium , specified in terms of the time - averaged transport coefficient @xmath84 .
this is confirmed by recent works ( see e.g. ref .
@xcite ) , showing that the inclusion of the medium evolution by a full hydrodynamical simulation does not significantly change or improve , with respect to the assumption of a static medium , the results for the suppression of the high-@xmath5 @xmath0 in central collisions at rhic .
we assume that our approach for the estimation of @xmath75 is valid for rapidities belonging to the `` central plateau '' of the charged particle distribution @xcite , which is expected to be as large as @xmath85 at lhc energies @xcite .
we now sample a @xmath72 value , according to the quenching weight @xmath86 , and we modify the quark kinematics to ( @xmath87 , @xmath88 ) . as in ref .
@xcite , heavy quarks that lose all their energy ( @xmath89 ) are redistributed according to the thermal distribution @xmath90 with @xmath91 .
the muons from the decay of these thermalized quarks populate the region @xmath92 , outside the range of interest for the present study .
we assume the heavy - quark rapidity to stay constant during the process of energy loss , since heavy quarks co - move with the longitudinally - expanding medium and any modification of the initial heavy - quark rapidity should remain small .
finally , we apply fragmentation and decay as described in section [ sec : hq ] . in ref .
@xcite the local @xmath24 transverse profile at central rapidity ( @xmath93 ) is assumed to be proportional to the density of binary collisions . since we want to study muon production in a broad rapidity range ( @xmath94 ) , we introduce a dependence of the local @xmath24 on the pseudo - rapidity @xmath95 in order to account for the reduced medium density in the forward direction .
namely , we assume the transport coefficient @xmath96 to scale as a function of pseudo - rapidity according to the gluon pseudo - rapidity density of the medium @xmath97 , this choice being justified by the fact that @xmath79 scales in @xmath95 according to @xmath97 .
we assume the pseudo - rapidity density of charged particles and the pseudo - rapidity density of gluons to have the same dependence on @xmath95 and we write @xcite : @xmath98\ , , \label{eq : qhatscale}\ ] ] where @xmath99 is a constant that sets the scale of @xmath24 .
we use two values of @xmath99 that were estimated for central collisions at @xmath23 by scaling the values extracted from an analysis of the light - flavour hadrons suppression at rhic @xcite .
the corresponding values for the parton - averaged ( @xmath100 ) and time - averaged ( @xmath101 ) transport coefficient , for simplicity . ] are @xmath102 and @xmath103 . for the pseudo - rapidity dependence we use the pseudo - rapidity distribution of charged particles predicted for collisions at the lhc in ref .
the effect of this dependence on the suppression of muons from heavy - quark decays is expected to be small , because @xmath104 is predicted to have only a modest variation in the range @xmath105 ( for illustration , for @xmath106 it would be reduced by about @xmath107 with respect to @xmath108 ) . before presenting our results , we point out that , in addition to radiative energy loss , there are other possible medium - induced effects ( e.g. : collisional energy loss of heavy quarks @xcite , in - medium hadronization and dissociation of heavy - flavour hadrons @xcite , formation of bound states and hadronization of heavy quarks via coalescence @xcite ) that could noticeably affect the muon spectrum for @xmath109 .
these effects have not been considered in the present study .
and one has to keep in mind that all current jet quenching models do not describe the suppression of single non - photonic electron @xmath5 spectra at rhic equally well as they do describe the suppression of pion spectra .
it is with this caveat that one should view the predictions for lhc based on these models .
we start by presenting , in the left - hand panel of fig .
[ fig:1 ] , the muon production cross - section as a function of transverse momentum in the 10% most central collisions at @xmath110 tev , when only nuclear shadowing is included .
the contributions from charm , beauty and weak gauge bosons are shown separately .
muons from charged pion and kaon decays and from drell - yan processes ( @xmath111 ) are ignored here , because both are expected to be negligible in the transverse momentum range @xmath112 @xcite . due to their large masses , w and z bosons
are mainly produced with small transverse momentum , @xmath113 .
therefore , the decay muons have typically @xmath114 .
the latest qualitatively explains the shape of the @xmath5 distribution of muons from the decay of the w and z , which `` peaks '' at @xmath115 . among the contributions that we compute , muons from charm decays are predominant in the low-@xmath5 range , 2@xmath22 . in the range 4@xmath116 beauty decays prevail , and at larger @xmath7 the w decays represent the largest contribution to the muon spectra .
we may note that these values for the `` crossing points '' in @xmath5 are somewhat dependent on the perturbative uncertainties of nlo calculations ( choice of the fragmentation and renormalization scales ) and on other systematics in the fragmentation and decay kinematics .
for example , a variation of @xmath117 of yield of the beauty component ( see section [ sec : hq ] ) implies a shift of the crossing point by approximately @xmath118 . as discussed in refs .
@xcite , the experimental measurement of the ratio of positive - to - negative muons could help to determine these crossing points .
for illustration of the expected significance for high-@xmath5 muons , we quote the production yield of muons from w decay in minimum - bias collisions at @xmath23 for an integrated luminosity of @xmath119 ( the expected integrated luminosity to be collected by each of the three experiments within one month of data - taking ) . in this case ,
about @xmath120 muons from w decays will be produced in the range @xmath121 .
out of them , about @xmath122 will be produced in @xmath123 and @xmath124 in @xmath125 . in view of exploring the final - state effects ( heavy - quark energy loss ) on the muon spectrum ,
we first focus on the influence of the initial - state effects , i.e. nuclear shadowing , by analysing the nuclear modification factor @xmath0 as a function of transverse momentum ( right - hand panel of fig .
[ fig:1 ] ) .
the short - dashed and dotted lines represent the heavy - quark decays contributions ; at low @xmath7 ( @xmath126@xmath22 ) they probe the small @xmath127 range of the gluon pdf @xmath128 , where , according to the eks98 parametrization @xcite , we have shadowing : the pdf in the pb nucleus is suppressed with respect to the pdf in the free proton ( @xmath129 ) .
thus , @xmath130 . for larger muon @xmath7 ( @xmath131 ) a higher @xmath127 range is probed , and we begin to explore the anti - shadowing region ( @xmath132 ) , so @xmath133 as is observed in fig .
[ fig:1 ] . for illustration , in table [ tab:1 ]
we report a qualitative estimation of the probed @xmath134 values and of the corresponding eks98 shadowing factors @xmath135 for heavy quarks and w / z bosons in collisions at @xmath39 tev , as a function of the decay - muon rapidity and transverse momentum ( @xmath136 is the mean value of @xmath137 ) .
the @xmath127 values were obtained using the pythia event generator for the leading order processes @xmath138 and @xmath139 .
for these two processes , we have , qualitatively , @xmath140 and @xmath141 , respectively ; and , for both , @xmath142 , @xmath143 .
weak gauge boson decays ( long - dashed line in fig .
[ fig:1 ] ) probe the quarks nuclear shadowing , which fluctuates around its mean value of @xmath144
. the overall nuclear modification factor of muons ( solid line in fig .
[ fig:1 ] ) increases rapidly with @xmath7 up to a value of about @xmath145 and then decreases to about @xmath144 . in order to explore the uncertainties due to the limited knowledge of the nuclear pdfs in the kinematic region relevant to our study , we considered two other parametrizations of the nuclear parton distribution : nds @xcite and hkn07 @xcite .
the values of the corresponding shadowing factors are reported in table [ tab:1 ] .
we note that the shadowing factor uncertainty for heavy quarks at high @xmath5 is 7% and for weak gauge boson is less than 5% . for heavy quarks at low @xmath5 ,
it becomes as large as 20% for charm and 10% for beauty .
note that the study of proton nucleus collisions at rhic energies has been absolutely necessary to disentangle cold and hot nuclear matter effects .
cold nuclear matter effects at lhc energies remain relatively unknown
. therefore , p pb runs at the lhc will be needed in order to fully understand the nuclear modification factors measured in pb pb . cccccccccccc & @xmath57 & @xmath7 & @xmath146 & @xmath147 & @xmath148 & @xmath149 & @xmath150 & @xmath151 & @xmath152 & @xmath153 + & 0 & 0 & & 0.50 & & 0.81 & & 0.74 & + & 0 & 30 & & 1.10 & & 1.09 & & 0.93 & + & 3 & 0 & @xmath154 & @xmath155 & 0.57 & & 0.83 & & 0.74 & + & 3 & 30 & @xmath156 & @xmath157 & 0.91 & & 0.96 & & 1.09 & + & 0 & 0 & & 0.77 & & 0.92 & & 0.83 & + & 0 & 30 & & 1.10 & & 1.00 & & 0.93 & + & 3 & 0 & @xmath158 & @xmath159 & 0.85 & & 0.93 & & 0.85 & + & 3 & 30 & @xmath156 & @xmath157 & 0.91 & & 0.96 & & 1.09 & + & 0 & all & & 0.89 & & 0.88 & & 0.83 & + & 3 & all & @xmath160 & @xmath161 & 0.76 & & 0.77 & & 0.85 & + & 0 & all & & 0.90 & & 0.89 & & 0.84 & + & 3 & all & @xmath160 & @xmath162 & 0.77 & & 0.76 & & 0.83 & + we now include in the calculation the in - medium energy loss for heavy quarks .
we start by considering the nuclear modification factor @xmath26 of muons from beauty decays in central ( 010% ) collisions , in order to study the effects of the b quark mass and of the dependence of the transport coefficient @xmath24 on @xmath95 according to @xmath163 .
the latter is relevant only in the large pseudo - rapidity range .
the result is shown in fig .
[ fig:2 ] .
the shaded band represents our baseline result for the @xmath24 range 25100 @xmath164 , with mass effect ( @xmath165 ) and with pseudorapidity dependence of @xmath24 .
the suppression obtained at high @xmath5 , where @xmath0 becomes independent of @xmath5 , is about a factor 5 ( 10 ) for @xmath166 .
the suppression for electrons from beauty decays was calculated within the same framework ( except for the treatment of hadronization and decay ) in ref .
@xcite , for central rapidity and @xmath167 . for the same @xmath5 range ,
we obtain similar values for @xmath0 .
with reference to fig .
[ fig:2 ] , by comparing the thick solid line ( @xmath165 ) and the long - dash line ( @xmath168 ) , we notice that the quark mass effect in parton energy loss increases @xmath0 by up to a factor of three for @xmath169 , and that some effect persists even at @xmath170 . when going to @xmath171 , the quark mass dependence becomes negligible since the quark mass becomes negligible with respect to the momentum . by comparing the thick and thin solid lines with @xmath172 , which are with and without @xmath95 dependence of @xmath24 respectively ,
we conclude that this dependence has basically no effect on the amount of suppression in the range @xmath4 .
figure [ fig:3 ] shows the muon @xmath5-differential cross section and the nuclear modification factor for muons from w , z , c and b decays in central ( 010% ) collisions at @xmath110 tev in the two pseudo - rapidity domains .
the results with transport coefficient values @xmath173 , 25 and 100 gev@xmath174/fm are reported , where @xmath175 corresponds to the case of no energy loss .
with reference to the upper panels of fig [ fig:3 ] , note that , since muons from w and z decays are unaffected by the energy loss , the crossing point in transverse momentum of the distributions of b - quark and w - boson decay muons shifts down by @xmath176 at large rapidities and by @xmath177 at mid - rapidity , when energy loss is included . concerning the muon @xmath26 ( lower panels of fig .
[ fig:3 ] ) , at large pseudo - rapidities ( on the left ) , with heavy - quark energy loss , the overall muon yield is suppressed by about a factor of 25 in the range @xmath178 , where the beauty contribution dominates . for higher @xmath5
, @xmath0 increases rapidly in the @xmath179 range and flattens at around 0.8 above @xmath180 .
the @xmath24-independence of the @xmath0 of overall muons at large @xmath7 is due to the fact that the w / z boson contribution to the yield becomes dominant ( see upper panels of fig .
[ fig:3 ] ) . at mid - rapidity
the behaviour is similar ( lower - right panel ) .
the small difference of the @xmath0 shape at different rapidities is due to the different proportion of heavy - quark and w / z boson decays .
for the same reason the nuclear modification factor without energy loss ( dot - dashed lines in the lower panels in fig .
[ fig:3 ] ) differs slightly from that obtained without acceptance cuts ( solid line in fig .
[ fig:1 ] ) . besides the -to - pp nuclear modification factor @xmath0 , also the central - to - peripheral nuclear modification factor
@xmath1 will provide information on the medium - induced suppression of b quarks .
@xmath1 is defined as : @xmath181 where the index @xmath182 stands for central ( peripheral ) collisions . from the experimental point of view
, the @xmath1 measurement will be more straight - forward than the @xmath0 measurement , for the following two reasons .
1 ) the measurements in pp and in will be affected by different systematic errors ( especially for the cross section normalization ) , which will add up in the @xmath0 uncertainty .
2 ) pp collisions at the lhc will have different c.m.s .
energy ( 14 tev ) with respect to ( 5.5 tev ) , therefore the muon spectra measured in pp will have to be extrapolated from 14 tev to 5.5 tev with the guidance of perturbative qcd calculations , introducing an additional systematic error of the order of 10% on @xmath0 @xcite . in our calculation , the initial - state effects are assumed to be the same in central and peripheral collisions namely , we do not include an impact parameter dependence for shadowing thus , they cancel out in the central - to - peripheral ratio . as a consequence
, the @xmath1 of muons from weak gauge boson decays is equal to one .
the central ( 010% ) to peripheral ( 4070% ) ratios are shown in fig .
[ fig:4 ] . in central ( 010% ) collisions the yield might be reduced with respect to peripheral collisions ( 4070% ) by a factor 23 in the @xmath7 range from about @xmath183 to about @xmath184 , where the b - quark contribution dominates .
when going to larger @xmath7 , the @xmath1 of muons increases fast and then flattens at around 0.8 at mid - rapidity and 1.0 at forward rapidity .
this difference at high @xmath5 between the two pseudo - rapidity regions is due to the different relative abundances of the heavy - quarks and weak bosons components . for the same reason ,
the curves for @xmath185 and @xmath103 cross each other at @xmath186 at large pseudo - rapidity and at @xmath187 at mid - rapidity .
we have checked that the uncertainties on the cross sections of muons from w decays and of muons from beauty decays ( approximately 10% and 30% , respectively , as discussed in section [ sec : baseline ] ) translate into a variation smaller than 5% of the @xmath0 and @xmath1 values for @xmath188 , while they have no effect at lower @xmath5 .
the effect of heavy - quark energy loss on the differential cross section of muons produced in and pp collisions at lhc energies has been investigated .
the most important contributions to the decay muon yield in the range @xmath189 have been included : b ( and c ) quarks have been computed using a nlo pqcd calculation supplemented with the bdmps mass - dependent quenching weights ; weak gauge bosons muonic decays have been computed using the pythia event generator . the heavy - quark mass - dependence of energy loss reduces the suppression of muon yields from beauty decays by about factor two for @xmath190 . to account for the decrease of the medium density at large pseudo - rapidity
, we assumed a decrease of the transport coefficient proportionally to @xmath104 , and we found that the effect on the muon @xmath5 distribution is negligible , especially at large transverse momentum .
we investigated the energy loss effect by means of the -to - pp and of the central - to - peripheral nuclear modification factors in the acceptance of the lhc experiments : atlas , alice , and cms . in the @xmath5 interval below approximately @xmath191 , where the beauty component is dominant , @xmath0 for 010% central collisions relative to pp and @xmath1 for 010% relative to 4070% collisions
are found to be about 0.20.4 and 0.30.5 , respectively .
then , we observe a steep rise in the beauty / w crossover interval 20@xmath192 , up to the values @xmath193 and @xmath194 in the interval above @xmath192 , dominated by w / z decay muons .
these muon nuclear modification factors could provide the first experimental observation of the b quark medium - induced suppression in collisions at the lhc .
the presence of a medium - blind component ( muons from w and z decays ) that dominates the high-@xmath5 muon yield will allow an intrinsic calibration of the medium - sensitive probe ( heavy quarks ) , because it will provide a handle on the strength of the initial - state effects that may alter the hard - scattering cross sections in nucleus
nucleus collisions at the unprecedented energies of the lhc .
the authors , members of the alice collaboration , would like to thank their alice colleagues for useful exchanges during the accomplishment of the present work .
in particular , we would like to thank f. antinori , p. crochet , a. morsch and j. schukraft for fruitful discussions , n. armesto and c. a. salgado for their helpful suggestions at the beginning of this work and the calculation of the hkn parameterized npdfs , and to s. kumano and r. sassot for the calculation of the hkn and the nds parameterized npdfs , and to m. mangano for the hvqmnr program implementation . +
this work is partly supported by the nsfc ( 10575044 and 10635020 ) , the key project of the chinese ministry of education ( 306022 and irt0624 ) and the france china particle physics laboratory , fcppl ( cnrs / in2p3 and chinese ministry of education ) . | we study the nuclear modification factors @xmath0 and @xmath1 of the high transverse momentum ( @xmath2 ) distribution of muons in collisions at lhc energies .
we consider two pseudo - rapidity ranges covered by the lhc experiments : @xmath3 and @xmath4 .
muons from semi - leptonic decays of heavy quarks ( c and b ) and from leptonic decays of weak gauge bosons ( w and z ) are the main contributions to the muon @xmath5 distribution above a few @xmath6 .
we compute the heavy quark contributions using available pqcd - based programs .
we include the nuclear shadowing modification of the parton distribution functions and the in - medium radiative energy loss for heavy quarks , using the mass - dependent bdmps quenching weights .
muons from w and z leptonic decays , that dominate the yield at high @xmath7 , can be used as a medium - blind reference to observe the medium - induced suppression of beauty quarks . , , , , and quark gluon plasma , relativistic heavy ion collisions , heavy quarks , weak gauge bosons , muons 24.85.+p , 25.75.dw , 25.75.-q |
let @xmath14 be a compact hamiltonian @xmath4-manifold with moment map , @xmath15 . if @xmath16 is finite , @xmath17 is a morse function , and its critical points , @xmath18 , are all of even index .
this has important consequences for the topology of @xmath2 : if we put an @xmath4-invariant riemannian metric , @xmath19 , on @xmath2 and let @xmath20 be the gradient vector field associated with @xmath19 and @xmath21 , then , for every critical point , @xmath18 , the unstable manifold at @xmath22 : @xmath23 supports a cohomology class , and these `` thom '' classes @xmath24 are an additive basis of the cohomology ring , @xmath25 . the same is true of the stable manifolds @xmath26 and their dual thom classes @xmath27 the main topic of this paper will be the symplectic version of what is sometimes called the multiplicative morse problem : given @xmath22 and @xmath28 in @xmath16 , @xmath29 can be expanded as a sum @xmath30 what are the @xmath31 s ? closely related to this is the question of determining the cohomology pairings : @xmath32 neither of these questions is easy to answer even when the structure of the cohomology ring itself is well understood .
for instance if @xmath2 is the coadjoint orbit of a compact lie group , the computation of the @xmath31 s is an important open problem in the theory of the schubert calculus and is the focus of a lot of recent activity .
( see , for instance , [ bgg ] , [ bh ] , [ bi ] , [ ko ] and [ kn ] . ) in this paper we will consider the _
equivariant _ version of this problem
. we will assume the action of @xmath4 on @xmath2 can be extended to a hamiltonian action of a torus , @xmath0 , of dimension @xmath33 , and replace the @xmath34 s in by their equivariant counterparts .
these equivariant thom classes generate @xmath35 as a module over the ring , @xmath36 , so one gets as above an identity @xmath37 but now the @xmath31 s are elements of the polynomial ring , @xmath10 .
when @xmath38 they are polynomials of degree zero ( _ i.e. _ real numbers ) and , in fact , coincide with the @xmath31 s in .
thus , they are in principle a much larger list of unknown quantities
. we will show , however , that they are in some sense , easier to compute due to the fact that , in equivariant cohomology , one has a much richer store of intersection invariants to play around with . more explicitly if @xmath39 and @xmath40 are submanifolds of @xmath2 and @xmath41 and @xmath42 their dual thom classes , the intersection number @xmath43 is zero except when @xmath39 and @xmath40 are of complementary dimension . on the other hand
if @xmath39 and @xmath40 are @xmath0-invariant and @xmath41 and @xmath42 their equivariant thom classes , the expression ( which is now an element of @xmath10 ) can be non - zero no matter what the relative dimensions of @xmath39 and @xmath40 are .
moreover , by the localization theorem of atiyah - bott - berline - vergne , , is a sum of local intersection invariants @xmath44 where @xmath45 is the quotient field of @xmath10 and @xmath22 a fixed point , and each of these is itself an intersection invariant . of particular interest for us will be certain intersection invariants of this type associated with the moment map , @xmath21 .
suppose that @xmath22 and @xmath28 are critical points of @xmath21 and that there are no critical values of @xmath21 in the interval @xmath46 .
let @xmath47 and let @xmath48 be the symplectic reduction of @xmath2 at @xmath49 . by the marsden - weinstein theorem , @xmath50 is a symplectic orbifold , and the action of @xmath0 on @xmath2 induces an action of the group @xmath51 on @xmath50 . the reduced spaces @xmath52 and @xmath53 are @xmath0-invariant symplectic sub - orbifolds of @xmath50 and so their equivariant intersection `` number '' @xmath54 is well - defined as an element of the subring , @xmath55 , of @xmath56 .
moreover , if @xmath57 is finite , then for every @xmath58 , the local intersection number @xmath59 is well - defined as an element of @xmath60 .
we will now describe the role of these intersection numbers in the computation of the @xmath31 s .
we will say that @xmath2 is _ a gkm manifold _
if , for all non - critical values , @xmath49 , @xmath57 is finite .
thus , being gkm is a necessary and sufficient condition for the invariants to be well - defined .
we recall some other characterizations of these manifolds .
[ th:1.1 ] @xmath2 is gkm if and only if , for every @xmath61 , the weights , @xmath62 , of the linear isotropy representation of @xmath0 on @xmath63 are pair - wise linearly independent ; i.e , @xmath64 is not a multiple of @xmath65 if @xmath66 . [ th:1.2 ] @xmath2 is gkm if and only if , for every codimension one subtorus , @xmath67 , of @xmath0 the connected components of @xmath68 are either fixed points of @xmath0 or imbedded copies of @xmath69 .
moreover , if a connected component , @xmath39 , is a copy of @xmath69 , the action of @xmath0 on @xmath39 is symplectomorphic to the standard action of @xmath70 on @xmath69 .
[ th:1.3 ] @xmath2 is gkm if and only if the one skeleton of @xmath2 @xmath71 is a finite union of embedded @xmath69 s .
theorem [ th:1.3 ] is an obvious consequence of theorem [ th:1.2 ] ; and it is easy to see that if the hypotheses of theorem [ th:1.2 ] hold , @xmath2 is gkm . for the other implications
see @xcite .
the intersection properties of the embedded @xmath69 s in the set can be described by an intersection graph , and a beautiful observation of goresky - kottwitz - macpherson is that one can read off the structure of the equivariant cohomology ring of @xmath2 from the `` action '' of @xmath0 on this graph .
more explicitly let @xmath72 be the graph whose vertices are the fixed points of @xmath0 and whose edges , @xmath73 , are copies , @xmath74 , of the @xmath69 s in theorem [ th:1.3 ] .
the graph structure on this collection of vertices and edges is given by defining the pair of vertices incident to an edge , @xmath73 , to be the set @xmath75 in particular , if @xmath73 is an _ oriented _ edge , its initial vertex , @xmath76 , is defined to be the `` south pole '' , @xmath22 , of the two - sphere , @xmath74 , and its terminal vertex , @xmath77 , to be the `` north pole '' , @xmath28 of @xmath74 .
the action of @xmath0 on the set can be described graph theoretically by two pieces of data : a function @xmath78 which assigns to each oriented edge , @xmath73 , of @xmath72 a one - dimensional representation , @xmath79 , of @xmath0 and a function , @xmath80 , which assigns to each vertex , @xmath22 , a @xmath81-dimensional representation , @xmath82 .
these functions are defined by letting @xmath79 be the representation of @xmath0 on @xmath83 , @xmath84 , and letting @xmath82 be the representation of @xmath0 on @xmath63 .
it is easily checked that @xmath78 and @xmath80 satisfy the axioms : @xmath85 where @xmath86 is the edge obtained from @xmath73 by reversing its orientation , @xmath87 is the kernel of @xmath79 and @xmath22 and @xmath28 are the vertices @xmath76 and @xmath77 . in particular , by , @xmath82 is determined by the @xmath88 s with @xmath89 .
since @xmath79 is a one - dimensional representation it is determined by its weight , @xmath90 ; so the `` action '' of @xmath0 on @xmath72 associated with @xmath78 and @xmath80 consists essentially of a labeling of the oriented edges , @xmath73 , of @xmath72 by weights , @xmath90 .
the axioms impose , of course , some condition on this labeling .
for instance is equivalent to @xmath91 now let @xmath92 be the equivariant cohomology ring of @xmath2 and @xmath93 the equivariant cohomology ring of @xmath3 . since @xmath3 is a finite disjoint union of fixed points and these fixed points are also the vertices , @xmath94 , of @xmath72 it follows that @xmath95 moreover , if @xmath96 is the inclusion , the map @xmath97 is injective , by a well - known result of kirwan .
the theorem of goresky - kottwitz - macpherson which we alluded to above asserts : [ th:1.4 ] a map @xmath98 is in the image of @xmath99 if and only if , for every edge , @xmath73 , of @xmath72 @xmath100 @xmath22 and @xmath28 being the vertices of @xmath73 and @xmath101 the annihilator of @xmath90 in @xmath102 .
this leads us to define the _ equivariant cohomology ring _ ,
@xmath103 of @xmath72 to be the set of all maps , @xmath104 satisfying .
each of the thom classes gets mapped by @xmath99 onto an element of @xmath103 and we will continue to use the notation , @xmath34 , for this `` combinatorial '' thom class . the main result of this article is a formula for this thom class as a kind of path integral over certain paths in @xmath72 .
before stating this result we ll describe a few basic properties of these combinatorial thom classes .
lets continue to denote by @xmath17 the restriction of the moment map , @xmath21 , to @xmath3 . identifying @xmath3 with @xmath94
, one can think of @xmath21 as a real - valued function on @xmath94 . by theorem [ th:1.2 ]
, @xmath21 takes on distinct values on the vertices @xmath76 and @xmath77 of an oriented edge , @xmath73 .
we will say that this edge is _ ascending _ if @xmath105 and _ descending _ if the reverse inequality is true .
more generally if @xmath106 is a path in @xmath72 we will say that @xmath106 is _ ascending _ if each of its edges is ascending . for every vertex , @xmath107 define the _ index _
, @xmath108 , of @xmath22 to be the number of descending edges , @xmath73 , with @xmath89 .
[ th:1.5 ] the thom class , @xmath34 , has the following properties : 1 .
its support is the set of all vertices of @xmath72 which can be joined to @xmath22 by an ascending path .
the value of @xmath34 at @xmath22 is @xmath109 the product , @xmath110 , being over all descending edges with @xmath89 . in certain instances
these properties uniquely characterize @xmath34 .
[ th:1.6 ] suppose that the indexing function , @xmath111 , @xmath112 , is strictly increasing along ascending paths .
then @xmath34 is the _
unique _ element of @xmath113 with properties 1 and 2 above .
we will now describe our `` path - integral '' formula for @xmath34 . as mentioned above this formula
will involve the hamiltonian action of the subgroup , @xmath4 , of @xmath0 on @xmath2 and the intersection invariants and .
let @xmath114 be the infinitesimal generator of this subgroup and let @xmath73 be an ascending edge of @xmath72 with @xmath115 . for any point , @xmath49 , on the interval between @xmath116 and @xmath117 , the @xmath4-reduced space , @xmath118 consists of a single point , @xmath119 .
let @xmath120 be the local intersection number .
[ th:1.7 ] for every @xmath121 @xmath122 the sum being over all ascending paths , @xmath106 , joining @xmath22 to @xmath28 , and the summands being defined by @xmath123 where @xmath124 are the edges of @xmath106 and @xmath125 * remarks : * 1 .
the local intersection number , @xmath120 , is equal to the global intersection number provided that there are no ascending paths in @xmath72 of length greater than one joining @xmath126 to @xmath127 . in particular , if @xmath106 is a longest path joining @xmath22 to @xmath28 all the intersection numbers in are global intersection numbers and in particular are elements of @xmath10 .
2 . in section [ sec:4 ]
we will give a purely combinatorial definition of @xmath120 .
as a corollary of theorem [ th:1.5 ] one gets for the formula @xmath128 summed over all configurations of paths , @xmath129 , @xmath130 and @xmath131 , where @xmath129 is an ascending path from @xmath22 to @xmath132 , @xmath130 an ascending path from @xmath28 to @xmath132 , @xmath131 a descending path from @xmath133 to @xmath132 and @xmath134 in particular : [ th:1.8 ] if the hypotheses of theorem [ th:1.6 ] are satisfied , @xmath135 and hence @xmath31 is equal to the sum .
a few words about the organization of this paper . in section [ sec:2 ]
we will give a brief account of the theory of @xmath0-actions on graphs ( based , for the most part , on material in @xcite ) .
in section [ sec:3 ] we will derive a preliminary version of theorem [ th:1.7 ] and then , in section [ sec:4 ] , deduce from it the version above , after first describing how to define the invariants combinatorially . in section [ sec:5 ]
we will attempt to demystify what is perhaps the most puzzling feature of the formula , the fact that all the summands are rational functions ( elements of the quotient field , @xmath136 ) , whereas the sum itself is a polynomial .
this indicates that a lot of mysterious cancellations are occurring in this summation ; and we will show _ how _ these cancellations occur in a few simple but enlightening examples . we would like to thank tara holm and sue tolman for helping us with the computations involved in these examples and allen knutson for pointing out to us antecedents in the combinatorial literature for formulas of type and .
let @xmath72 be a finite @xmath81-valent graph .
given an oriented edge , @xmath73 , of @xmath72 we will denote by @xmath76 the initial vertex of @xmath73 and by @xmath77 the terminal vertex ; and we will denote by @xmath86 the edge obtained from @xmath73 by reversing its orientation . thus @xmath137 and @xmath138 .
[ dfn:2.1 ] let @xmath78 be a function which assigns to each oriented edge , @xmath73 , of @xmath72 a one dimensional representation , @xmath139 ; and let @xmath80 be a function which assigns to each vertex , @xmath22 , a @xmath81-dimensional representation of @xmath140 .
we will say that @xmath78 and @xmath80 define _ an action of @xmath0 on @xmath72 _ if the axioms are satisfied .
let @xmath90 be the weight of the representation @xmath79 . by
, the weights , @xmath141 , determine the representation @xmath82 up to isomorphism ; so the action of @xmath0 on @xmath72 is basically just a labeling of the edges of @xmath72 by weights .
we will denote the function , @xmath142 , by @xmath143 and call it the _ axial function _ of the action of @xmath0 on @xmath72 .
the axioms can be reformulated as statements about @xmath143 : [ prop:2.2 ] axiom is satisfied iff @xmath144 and axiom is satisfied iff one can order the weights @xmath145 so that @xmath146 ( we will leave the proof of these assertions as an easy exercise . ) [ dfn:2.3 ] the action of @xmath0 on @xmath72 is a _ gkm action _ if , for all vertices , @xmath22 , the weights , @xmath147 , are pair - wise linear independent . from now on we will assume ,
unless we state otherwise , that the action of @xmath0 on @xmath72 has this property . for every vertex , @xmath22 , of @xmath72 let @xmath148 be the set of oriented edges , @xmath73 , with @xmath89 .
[ dfn:2.4 ] a _ connection _ on @xmath72 is a function which assigns to every oriented edge , @xmath73 , a bijective map @xmath149 satisfying @xmath150 .
this connection is _ compatible with the action of _
@xmath0 if , for every oriented edge , @xmath73 , with @xmath89 and every edge , @xmath151 @xmath152 thus the existence of a @xmath0-compatible connection is a slight sharpening of the identity .
it is easy to see that @xmath0-compatible connections exist , and we will assume henceforth that @xmath72 is equipped with such a connection . let @xmath94 be the set of vertices of @xmath72 and @xmath153 the set of oriented edges . motivated by the theorem of goresky kottwitz macpherson we define the equivariant cohomology ring , @xmath154 , of @xmath72 to be the set of maps , @xmath155 , satisfying the compatibility condition for all @xmath156 .
this ring has a natural grading the degree @xmath157 .
( it is , however , more natural in this algebraic context . ) ] @xmath158 and contains @xmath10 as a subring : the ring of constant maps of @xmath94 into @xmath10 .
the proof of theorem [ th:1.7 ] will require a number of results about the structure of @xmath154 as an @xmath10 module .
these results were proved in an earlier paper of ours on `` equivariant morse theory on graphs '' ( @xcite ) , and we will refer to this paper for a detailed treatment of the material in the next few paragraphs .
let @xmath159 given an element , @xmath160 , we will say that an oriented edge , @xmath73 , is _ ascending with respect to _ @xmath161 if @xmath162 . for every vertex , @xmath22 , let @xmath108 , the _ index _ of @xmath22 , be the number of ascending edges , @xmath73 , with @xmath163 . [ dfn:2.5 ] the @xmath164 betti number , @xmath165 , is the number of vertices , @xmath22 , of @xmath72 for which @xmath166 .
* remark : * the definition of @xmath108 depends upon the choice of @xmath161 but @xmath167 turns out not to .
( see ( * ? ? ?
* theorem 2.6 ) ) .
a function @xmath168 is a ( @xmath161-compatible ) _ morse _ function if , for every ascending edge @xmath73 , @xmath169 .
it is not obvious , and in fact not true , that morse functions exist .
a necessary and sufficient condition for the existence of a morse function is that , for every ascending path in @xmath72 the initial vertex of this path is distinct from its terminal vertex ( _ i.e. _ , there are no ascending `` loops '' ) .
if a morse function exists , however , one can easily perturb it so that it is _ injective _ as a map of @xmath94 into @xmath170 . from now on we will let @xmath21 be a fixed morse function with this property .
the topological results discussed in section [ sec:1 ] prompt one to make the following morse - theoretic conjectures about the equivariant cohomology ring of a graph .
[ cnj:1 ] @xmath171 is a free @xmath10 module with @xmath172 generators of degree @xmath173 .
[ cnj:2 ] @xmath171 is freely generated as an @xmath174 module by a family of `` thom classes ''
@xmath175 satisfying @xmath176 and @xmath177 @xmath178 being the set of vertices which can be joined to @xmath22 by an ascending path and @xmath179 being the set of descending edges in @xmath148 .
it is clear that conjecture [ cnj:2 ] implies conjecture [ cnj:1 ] , and it is not difficult to prove that conjecture [ cnj:1 ] implies conjecture [ cnj:2 ] ( see @xcite ) .
therefore , since conjecture [ cnj:1 ] does nt depend on the choice of an orientation of @xmath72 ( _ i.e. _ the choice of a polarizing vector @xmath160 ) , the same is true of conjecture [ cnj:2 ] . in particular ,
if we reverse the orientation ( replace @xmath161 by @xmath180 ) , we get from conjecture [ cnj:2 ] the existence of thom classes , @xmath181 , @xmath182 , associated with the morse function @xmath183 .
unfortunately these conjectures are not true in general ; however there is a useful necessary and sufficient condition for them to be true involving certain subgraphs of @xmath72 .
[ dfn:2.6 ] a subgraph , @xmath184 , of @xmath72 is _ totally geodesic _ if , for every pair of edges , @xmath73 and @xmath185 , of @xmath184 , with @xmath186 , @xmath187 is also an edge of @xmath184 . note that
if @xmath184 is a totally geodesic subgraph of @xmath72 the restriction to it of @xmath143 is , by , an axial function on @xmath184 ; so each of these subgraphs is equipped with an action of @xmath0 .
an important example of a totally geodesic subgraph is the following .
let @xmath188 be a subspace of @xmath189 , and let @xmath190 be the subgraph whose edges are the set @xmath191 ( it is clear , by and that this _ is _ totally geodesic . ) one of the main results of [ gz3 ] is the following .
[ th:2.7 ] conjecture [ cnj:2 ] is true for @xmath72 if and only if , for every two - dimensional subspace , @xmath188 , of @xmath189 , conjecture [ cnj:2 ] is true for @xmath190 .
thus , to verify that conjecture 2 holds for @xmath72 it suffices to verify it for these subgraphs ( which is usually much easier than verifying it for @xmath72 itself ) .
the proof of theorem [ th:2.7 ] involves a graph - theoretic version of symplectic reduction .
we will say that @xmath49 is a _ critical value _ of the morse function @xmath192 if @xmath193 for some @xmath182 and , otherwise , @xmath49 is a _
regular value_. let @xmath49 be a regular value of @xmath21 and let @xmath194 be the set of oriented edges , @xmath73 , of @xmath72 with @xmath195 .
we show in @xcite that @xmath194 is the set of vertices of a hypergraph , @xmath196 .
thus the elements of @xmath194 are both edges of the graph @xmath72 and vertices of this hypergraph .
it is useful to distinguish between the two roles they play by saying that `` an edge , @xmath73 , intersects @xmath196 in a vertex , @xmath197 . ''
let @xmath198 be the annihilator of @xmath161 in @xmath189 .
for each oriented edge , @xmath73 , of @xmath72 we define a map @xmath199 by setting @xmath200 this extends to a ring homomorphism @xmath201 and , from , we get a ring homomorphism @xmath202 by setting @xmath203 ( the two terms on the right are equal by . )
we show in @xcite that @xmath204 maps @xmath205 into the cohomology ring , @xmath206 , of the hypergraph @xmath196 .
we wo nt bother to review here the definition of this hypergraph cohomology ring ( which is quite tricky ) since one of the main theorems of @xcite asserts that , if the hypotheses of theorem [ th:2.7 ] hold and if @xmath161 satisfies a certain genericity condition ( which we will spell out below ) , the map @xmath207 is a submersion .
hence , thanks to this theorem , one can _ define _ @xmath208 to be the image of @xmath204 .
a key step in the proof of theorem [ th:2.7 ] is a theorem which describes how the structure of the ring @xmath209 changes as one passes through a critical value of @xmath49 .
more explicitly suppose @xmath49 and @xmath210 are regular values of @xmath21 and suppose that there exists a unique vertex , @xmath22 , with @xmath211 .
in addition suppose that , for @xmath212 , @xmath213 , @xmath214 , @xmath215 @xmath216 except when the two sides of are forced to be equal ( _ i.e. _ , except when @xmath217 and @xmath218 or @xmath219 and @xmath220 ) .
the inequality is unfortunately not satisfied for all elements , @xmath161 , of the set , but one can show that those @xmath161 s for which it is satisfied form an open dense subset of this set .
let @xmath133 be the index of @xmath22 and let @xmath221 .
let @xmath222 be the descending edges in @xmath148 and @xmath223 be the ascending edges in @xmath148 .
let @xmath224 then @xmath225 is a subset of @xmath194 , @xmath226 a subset of @xmath194 and @xmath227 where @xmath228 is the intersection of @xmath194 and @xmath229 .
let @xmath230 then one has projection maps @xmath231 and , from these projection maps , pull - back maps , @xmath232 and @xmath233 , embedding the rings @xmath234 and @xmath235 into the ring @xmath236 moreover the ring @xmath237 sits in the ring as the set of maps @xmath238 supported on @xmath239 , and the ring @xmath240 sits inside the ring ; so _ all _ the rings can be regarded as subrings of .
let @xmath241 be a basis of @xmath198 and @xmath242 a fixed element of @xmath189 with @xmath243 .
let @xmath244 with @xmath245 and with the @xmath246 s in @xmath198 .
consider the maps @xmath247 the first two of these maps depend on the choice of @xmath242 ; however , @xmath248 is intrinsically defined since @xmath249 is just @xmath250 also , by the genericity condition the map , @xmath248 sends @xmath251 injectively into @xmath198 and , as a consequence , @xmath252 and @xmath253 map @xmath254 and @xmath255 injectively into @xmath198 .
define the cohomology ring , @xmath256 , to be the set of all maps of @xmath225 into @xmath257 of the form @xmath258 and define @xmath259 to be the set of all maps of @xmath255 into @xmath257 of the form @xmath260 the theorem we alluded to above asserts [ th:2.8 ] for every @xmath261 and every @xmath262 , @xmath263 , there exists a unique @xmath264 and unique @xmath265 , @xmath266 such that @xmath267 * remarks : * 1 .
this theorem gives one a concrete picture of how @xmath268 changes as one goes through a critical point of @xmath21 .
namely it shows that @xmath269 can be obtained from @xmath268 by a `` blow - up '' followed by a `` blow - down '' .
( compare with ( * ? ? ?
* theorem 2.3.2 ) . ) 2 .
this theorem can also be used to map cohomology classes in @xmath268 into cohomology classes in @xmath270 . by setting @xmath271 , @xmath272 , in one
gets from the cohomology class @xmath273 a unique cohomology class @xmath274 .
( this observation will be heavily exploited in the next section . )
an important ingredient in the proof of theorem [ th:2.8 ] is the following .
[ th:2.9 ] if @xmath275 is in @xmath276 , then its restriction to @xmath225 is in @xmath277 .
( it is in the proof of this result that the hypotheses of theorem [ th:2.7 ] are needed . )
we will describe in this section how to compute the combinatorial thom class , @xmath278 , at an arbitrary point @xmath22 on the flow - up @xmath279 .
we recall that @xmath278 is canonically defined _ only _ if the index function @xmath280 , is strictly increasing along ascending paths in @xmath72 . assuming that @xmath281 has this property
, we will show below that there is a simple inductive method for computing @xmath278 on a critical level , @xmath49 , of @xmath21 if one knows the values of @xmath278 on lower critical levels .
then , later in this section , we will show that this method works even when the hypothesis about @xmath281 is dropped .
let @xmath282 and @xmath283 .
the first step in this induction is to set @xmath284 for all vertices , @xmath28 , with @xmath285 and set @xmath286 equal to @xmath287 , as in .
now let @xmath288 and suppose , by induction , that @xmath289 is defined for all @xmath28 with @xmath290 and is zero unless @xmath28 is in @xmath291 .
let @xmath22 be a vertex with @xmath292 .
let @xmath293 and let @xmath294 , @xmath295 , be the descending edges in @xmath72 with @xmath296 .
then the vertices , @xmath297 , are points where @xmath278 is already defined .
we will prove below .
[ lem:3.1 ] there exists a unique polynomial , @xmath298 such that @xmath299 * remark : * the `` uniqueness '' part of this lemma is where the hypothesis on @xmath281 is used .
if @xmath300 and @xmath301 hence , if @xmath302 , @xmath275 is identically zero . using this lemma , set @xmath303 , and continue with the induction until the set of vertices of @xmath72 is exhausted .
it is clear from that this construction gives us a map : @xmath304 satisfying and that this map is supported on @xmath178 . by giving a constructive proof of the `` existence '' part of lemma [ lem:3.1 ] the induction argument we just sketched can be converted into a formula for @xmath278 , and this will be the main goal of this sections .
note that the solution of is basically an interpolation problem : finding a polynomial with prescribed values at @xmath305 . to solve this problem
constructively , we review a few elementary facts about `` interpolation '' .
the basic problem in interpolation theory is to find a polynomial @xmath306 which takes prescribed values @xmath307 at @xmath308 distinct points , @xmath309 , on the complex line .
the solution of this problem is more or less trivial .
the polynomial @xmath310 satisfies and is the only polynomial of degree less than @xmath308 which _ does _ satisfy .
moreover , from one gets an explicit formula for the @xmath311 s in .
let @xmath312 where @xmath313 is the @xmath133-th elementary symmetric function in the variables , @xmath314 , @xmath315 .
then by @xmath316 one consequence of is an inversion formula for the vandermonde matrix , @xmath317 , with entries @xmath318 if @xmath319 then by and : @xmath320 in particular @xmath321 it is sometimes convenient to write the inversion formula in terms of the elementary symmetric functions @xmath322 rather than in terms of the @xmath313 s .
to do so , we note that @xmath323 ( to derive observe that @xmath324 and compare coefficients of @xmath325 on both sides . )
substituting into one gets an alternative inversion formula for the vandermonde matrix @xmath326 finally we note a couple of trivial consequences of and . from and
the identity @xmath327 we conclude that the sum @xmath328 is zero if @xmath173 is less than @xmath308 and @xmath329 if @xmath330 ; and from and the identity @xmath331 we conclude that @xmath332 in the applications which we will make of these identities below the @xmath309 s will be indeterminants and the @xmath333 s polynomials in these indeterminants , and we will want to know when the @xmath311 s are also polynomials in these indeterminants . to answer this question
we will show that these identities have a simple `` topological '' interpretation : suppose one is given a graph , @xmath72 , and an action of @xmath0 on @xmath72 defined by an axial function , @xmath334 .
one of the main results of an earlier paper of ours is that there is a canonical integration operation @xmath335 defined by @xmath336 where @xmath337 ( see @xcite .
this formula is the formal analogue of the standard localization theorem @xcite-@xcite in equivariant derham theory . )
in particular let @xmath338 be the complete graph on @xmath308 vertices .
denote these vertices by @xmath339 , and let @xmath340 be a basis of @xmath189 .
it is easy to check that the map @xmath341 which assigns the weight @xmath342 to the edge joining @xmath343 to @xmath344 , is an axial function , and that the map @xmath345 is an element of @xmath346 .
we claim that @xmath347 generate @xmath348 as a module over @xmath10
. to see this let @xmath349 be the cohomology class @xmath350 then simply asserts that @xmath351 in particular if @xmath275 is any cohomology class , then one can express @xmath275 as a sum @xmath352 where @xmath353 this proves the assertion : [ prop:3.2 ] if @xmath354 are polynomials in @xmath355 and the function @xmath356 solves the interpolation problem @xmath357 then the @xmath311 s are polynomials in @xmath355 if and only if @xmath342 divides @xmath358 .
if @xmath342 divides @xmath358 the map @xmath359 is in @xmath360 .
let s come back now to theorem [ th:2.8 ] and the application of it which we discussed at the end of section [ sec:2 ] .
as in theorem [ th:2.8 ] let @xmath49 and @xmath210 be regular values of @xmath21 , and suppose that there is just one vertex , @xmath22 , of @xmath72 with @xmath361 . by setting @xmath362 in one gets a map @xmath363 sending @xmath364 to @xmath365 , and by the results above one can give a fairly concrete description of this map .
let s order the edges @xmath366 so that @xmath367 are descending and @xmath368 are ascending , and let @xmath225 and @xmath226 be the vertices of @xmath196 and @xmath369 corresponding to the @xmath370 s , @xmath371 , and the @xmath372 s , @xmath373 . then @xmath374 @xmath228 being the vertices which are common to @xmath196 and @xmath375 . to simplify notation we will identify @xmath225 with the set @xmath376 and @xmath255 with the set @xmath377 .
let @xmath364 be in @xmath378 and let @xmath379 .
then , by and @xmath380 for @xmath381 in @xmath255 and @xmath344 and @xmath173 in @xmath225 ; and @xmath382 the identity has the following simple interpretation .
let @xmath383 then , by , @xmath384 solves the interpolation problem @xmath385 on the other hand , by theorem [ th:2.9 ] @xmath386 so , by proposition [ prop:3.2 ] , @xmath384 is a polynomial in @xmath387 and hence also a polynomial in @xmath388 , _
i.e. _ , an element of the ring , @xmath10 .
in fact , if @xmath389 and @xmath390 , @xmath384 is the _
unique _ element of @xmath391 satisfying .
now by , @xmath392 , so simply says that @xmath393 thus to summarize , we have proved : [ th:3.3 ] the map @xmath394 is the identity map on @xmath228 and on @xmath225 is the `` flip flop '' @xmath395 since @xmath194 and @xmath229 are finite sets , the map @xmath396 is defined by a matrix with entries @xmath397 an important property of this matrix is the markov property : @xmath398 it suffices to check this for @xmath399 , _
i.e. _ it suffices to check that @xmath400 however , by , this sum is equal to @xmath401 which is equal to @xmath329 by , with @xmath402 .
we will next give a more intrinsic description of @xmath396 and of the polynomial @xmath403 in .
we recall that @xmath404 hence @xmath405 and therefore @xmath406 where @xmath79 is the map .
similarly the polynomial @xmath403 is just @xmath407 by iterating we will extend the definition of @xmath396 to arbitrary regular values of @xmath21 with @xmath408 .
let @xmath409 be regular values of @xmath21 with @xmath410 and @xmath411 such that there exists a unique vertex , @xmath412 , with @xmath413 .
let @xmath414 and let @xmath415 be the map @xmath416 we will list a few properties of this map : * 1 ) . *
this map is defined by a matrix with entries @xmath417 and since all the factors on the right hand side of have the markov property , this matrix also has this property . *
* the matrix version of asserts that @xmath418 summed over all sequences @xmath419 with @xmath420 . by and , a large number of the matrix entries in this formula are either @xmath329 or @xmath421 : if @xmath422 is an ascending edge which intersects @xmath423 in @xmath424 and @xmath425 in @xmath426 , then @xmath427 this fact can be exploited to write the sum above more succinctly . for every pair of edges , @xmath73 and @xmath185 , with @xmath428 , let @xmath367 be the descending edges in @xmath148 , ordered so that @xmath429 and let @xmath430 then @xmath431 can be written as a weighted sum : @xmath432 over all ascending paths , @xmath106 , in @xmath72 whose initial edge intersects @xmath196 in @xmath433 and whose terminal edge intersects @xmath375 in @xmath434 , the weighting of the path , @xmath106 , being given by @xmath435 where the @xmath436 s are the edges of @xmath106 , ordered so that for @xmath437 , @xmath438 . *
* the map can also be viewed as a series of `` flip flops '' .
let @xmath364 be an element of @xmath439 and let @xmath440 .
then @xmath441 maps @xmath442 to @xmath333 by a map of the form .
let s denote the polynomial , @xmath403 , in by @xmath443 .
we claim : [ prop:3.4 ] if @xmath412 is joined to @xmath444 by an ascending edge , @xmath73 , @xmath445 this is equivalent to asserting that @xmath446 however , is , by definition , the common value of @xmath447 , @xmath448 , at the vertices , @xmath424 , at which @xmath73 intersects @xmath423 . *
in particular let @xmath449 be an arbitrary vertex of @xmath72 ; and choose @xmath49 and @xmath210 such that there are no critical values of @xmath21 on the interval , @xmath450 and such that @xmath451 , @xmath182 .
order the edges @xmath452 in @xmath453 so that @xmath367 are descending and @xmath454 are ascending . for @xmath455
let @xmath456 be the vertex at which @xmath372 intersects @xmath196 and let @xmath457 be the map defined by @xmath458 [ prop:3.5 ] @xmath364 is an element of @xmath268 . by , @xmath459 .
( this proof assumes that there _ exists _ a thom class , @xmath278 , having the properties listed in theorem [ th:1.5 ] .
alternatively , proposition [ prop:3.5 ] can be proved directly using a more sophisticated definition of @xmath460 than that which we gave in section [ sec:2 ] . for more details
see @xcite . ) by applying the sequence of flip flops , @xmath441 , to the @xmath364 above , we get a polynomial , @xmath461 for each vertex , @xmath412 , of @xmath72 with @xmath462 . on the other hand , we can define @xmath463 for @xmath464 to be equal to at @xmath449 and equal to zero otherwise . by @xmath463
satisfies the cocycle condition at all vertices except @xmath449 , and by it satisfies this condition at @xmath449 as well .
thus , if the index function , @xmath465 , is strictly increasing along ascending paths , this settles the existence part of lemma [ lem:3.1 ] and justifies the induction method for constructing @xmath466 which we outlined at the beginning of this section .
on the other hand , if @xmath281 fails to satisfy this hypothesis , the assignment , @xmath467 , still defines an element of @xmath468 with the properties listed in theorem [ th:1.5 ] ; however , it wo nt be the only element with these properties and may not even be the optimal element with these properties . *
* from one gets the following `` path integral '' formula for @xmath34 .
if @xmath73 is an ascending edge of @xmath72 , let @xmath469 and let @xmath470 be the descending edges in @xmath148 , ordered so that @xmath429
. let @xmath471 then by , and @xmath472 summed over all ascending paths in @xmath72 joining @xmath449 to @xmath22 , @xmath473 being defined by @xmath474 where @xmath212 is the initial edge of @xmath106 and @xmath475 is the terminal edge of @xmath106 .
we will show below how to recast the formula into the form and will also show that , if the hypothesis of theorem [ th:1.6 ] is satisfied , one can deduce from the formula that we described in section 1 for the products of thom classes .
first , however , we will examine this hypothesis in more detail : suppose the graph @xmath72 is connected and admits a family of thom classes , @xmath7 , @xmath182 , which generates @xmath113 as a free module over the ring @xmath10 , and have the properties and . by @xmath476 hence there is a unique vertex , @xmath449 , with @xmath477 .
let @xmath22 be an arbitrary vertex of @xmath72 and let @xmath106 be an ascending path with terminal endpoint @xmath22 .
if @xmath106 is of _ maximal _ length , its initial vertex has to be @xmath449 , since every other vertex has a descending edge .
let @xmath116 be the length of this longest path .
if @xmath22 can be joined to @xmath28 by an ascending edge , @xmath116 is strictly less than @xmath117 , so the map @xmath478 is a morse function .
[ th:4.1 ] the index function , @xmath281 , is strictly increasing along ascending paths if and only if @xmath479 ( _ i.e. _ if and only if the morse function , @xmath21 , is self - indexing . ) moreover ,
if @xmath21 has this property then , for every pair of vertices , @xmath182 and @xmath480 , the length of the longest ascending path from @xmath22 to @xmath28 is @xmath481 .
it suffices to prove the last assertion , and it suffices by induction to prove this assertion for paths of length one .
this we will do by proving a slightly stronger assertion .
[ th:4.2 ] let @xmath73 be an ascending edge joining @xmath22 to @xmath28 .
if @xmath73 is the only ascending path from @xmath22 to @xmath28 then @xmath482 let @xmath483 be the totally geodesic subgraph of @xmath72 consisting of the single edge , @xmath73 , and vertices @xmath22 and @xmath28 .
the thom class , @xmath484 , of @xmath483 is defined by @xmath485 and @xmath486 it is easily checked that @xmath487 . a cohomology class ,
@xmath488 is supported on @xmath489 iff @xmath490 , @xmath491 .
suppose now that @xmath73 satisfies the hypotheses of theorem [ th:4.2 ]
. then @xmath492 is supported on @xmath489 ; so , by the lemma , @xmath493 in particular @xmath494 so @xmath495 . coming back to the formula
lets first consider the simplest summands in this formula , those associated with paths , @xmath106 , of length one . for each @xmath121 denote by @xmath496 and @xmath497 the descending and ascending edges in @xmath498 and let @xmath499 be defined as in .
let @xmath106 be an ascending path of length one consisting of a single edge , @xmath73 , with @xmath500 and @xmath501 .
then by and @xmath502 where @xmath110 in the enumerator is a product over the edges @xmath503 and @xmath504 in the denominator is the product over the edges @xmath505 .
let @xmath506 be the connection along this edge and let @xmath507 .
we define @xmath508 note that @xmath509 restricts to a bijection @xmath510 if @xmath511 , then implies @xmath512 therefore if @xmath513 , then the terms corresponding to @xmath436 and @xmath514 in cancel each other and we obtain @xmath515 where @xmath516 and @xmath517 if @xmath106 is the only ascending path from @xmath22 to @xmath28 , then @xmath518 has an interpretation as an `` intersection number '' : by the quotient , @xmath519 is an element of @xmath520
. let @xmath49 be a point on the interval @xmath521 and let @xmath197 be the vertex of @xmath196 corresponding to @xmath73 .
if we apply the kirwan map @xmath522 to this quotient and evaluate at @xmath197 we get an element of @xmath523 .
we claim @xmath524 a direct computation shows that @xmath525 hence , follows from .
we will now show that the right hand side of can be interpreted as a `` pairing '' of the cohomology classes @xmath526 and @xmath527 .
we pointed out in section [ sec:3 ] that the localization formula in equivariant derham theory enables one to define an integration operation on @xmath113 .
the analogue of this result for @xmath196 asserts that there is an integration operation @xmath528 mapping @xmath529 to the sum @xmath530 where @xmath531 @xmath73 being the edge of @xmath72 which intersects @xmath196 of the vertex @xmath532 . in particular ,
consider the product in @xmath276 of @xmath526 and @xmath527 .
if @xmath73 is the only ascending path in @xmath72 joining @xmath22 to @xmath28 this product is zero except at the point @xmath197 ; so by @xmath533 which is the formal analogue of the intersection number .
* remarks : * 1 . by theorem [ th:4.2 ] , @xmath495 .
one can see by inspection that the right hand side of , which is by definition an element of @xmath523 , is of degree @xmath534 .
in particular , if @xmath281 is a self - indexing morse function , the right hand side of is just a constant .
if the edge , @xmath73 , is not the only path joining @xmath22 to @xmath28 , the identity is still true ; however the right hand side of is in @xmath535 and has to be interpreted as the formal analogue of the local intersection number .
we now return to the general case .
let @xmath536 be an ascending path from @xmath22 to @xmath28 , let @xmath537 be an ascending path from @xmath28 to @xmath133 , and let @xmath538 be the ascending path from @xmath22 to @xmath133 obtained by joining @xmath539 and @xmath540 .
a direct computation shows that @xmath541 where @xmath436 is the last edge of @xmath539 and @xmath372 is the first edge of @xmath540 , both pointing upward .
let @xmath542 be an ascending path .
we will express the contribution @xmath473 by breaking up the path @xmath106 into its constituent edges .
then @xmath543 therefore the contribution @xmath473 of the path @xmath106 is @xmath544 in view of we can also write this in the form , @xmath436 being the edge of @xmath72 joining @xmath545 to @xmath412 and @xmath120 being the local intersection number .
if we reverse the orientation of @xmath72 replacing @xmath161 with @xmath180 and the morse function @xmath21 by @xmath183 , we get a formula similar to for @xmath181 @xmath546 the sum being over _ descending _ paths from @xmath22 to @xmath28 . moreover , the @xmath473 s in are easy to compute in terms of the @xmath473 s in .
to see this lets consider as above the simplest example of an ascending path in @xmath72 , an ascending edge , @xmath73 , joining @xmath22 to @xmath28 . by and @xmath547 so ,
by and @xmath548 now let @xmath106 be an ascending path of length @xmath549 from @xmath22 to @xmath28 and let @xmath550 be the same path traced in the reverse direction
. then by and @xmath551 we are now finally in position to compute the cohomology pairing . by , and the integral @xmath552 is equal to the sum @xmath553 summed over all triples @xmath129 , @xmath130 , @xmath131 consisting of an ascending path , @xmath129 , from @xmath22 to @xmath132 , an ascending path , @xmath130 , from @xmath28 to @xmath132 , and a descending path , @xmath131 , from @xmath133 to @xmath132 .
( see figure [ fig : configuration ] . ) thus , in particular , if there exist no such configurations , @xmath554 .
now suppose that the hypothesis of theorem [ th:1.6 ] is satisfied , _ i.e. _ @xmath281 is a self - indexing morse function .
then we claim that @xmath555 in fact if @xmath556 the supports of @xmath7 and @xmath557 are non - overlapping so is automatically zero ; and if @xmath558 , then the support of @xmath559 consists of the single point @xmath22 and it is easy to verify that is equal to one .
thus is trivially true except when @xmath480 and @xmath560 . in this case
however , @xmath561 so @xmath562 so the integral is zero just by degree considerations .
thus if we substitute the sum @xmath563 for @xmath564 in we obtain for @xmath565 the formula .
each of the summands in is a rational function : an element of the quotient field , @xmath45 ; however , the sum itself is a polynomial , so the singularities in the individual summands are mysteriously cancelling each other out
. we will discuss below a few simple examples in which one can see how some of these cancellations are happening .
suppose @xmath106 is a longest ascending path from @xmath22 to @xmath28 .
let @xmath566 be the edges of @xmath106 ordered so that @xmath567
. then @xmath568 is the _ only _ path joining @xmath569 to @xmath570 ; hence the intersection numbers , @xmath571 , are all global intersection numbers of the form and are in @xmath10 .
hence the factor @xmath572 in the formula is in @xmath10 .
if , in addition , the morse function @xmath21 is self - indexing , this factor is a polynomial of degree zero , _
i.e. _ is just a constant .
let @xmath106 be the path consisting of the single edge , @xmath73 , joining @xmath22 to @xmath28 and let @xmath129 be the path @xmath573 .
assume @xmath129 is a longest path joining @xmath22 to @xmath28 and that @xmath574 .
we claim that @xmath575 hence @xmath581 however , by , @xmath582 ; hence we can rewrite this as @xmath583 so @xmath584 is equal to the expression : @xmath585 however , @xmath586 so the term in parentheses can be rewritten @xmath587 graph theoretically , the flag variety @xmath589 is the permutahedron : a cayley graph associated with the weyl group of @xmath590 , the symmetric group @xmath591 .
each vertex of this graph corresponds to a permutation @xmath592 , and two permutations @xmath593 and @xmath594 , are adjacent in @xmath72 if and only if there exists a transposition @xmath595 , @xmath596 with @xmath597 .
moreover , if @xmath73 is the edge joining @xmath593 to @xmath598 , the weight labeling @xmath73 is @xmath599 where @xmath600 is the standard basis vectors of the lattice @xmath601 . the connection
@xmath509 along this edge is given by @xmath602 the permutahedron is a bi - partite graph , with the two sets of vertices corresponding to even , respective odd permutations . in the special case @xmath606 ,
this graph is a _
complete _ bi - partite graph , and the corresponding labeling is shown in figure [ fig : flag ] .
suppose the graph @xmath72 is connected and hence has a unique vertex , @xmath449 , of index zero .
then the thom class , @xmath618 , is the unique generator of @xmath619 , with @xmath620 .
thus @xmath621 for all vertices @xmath22 .
we will show how to deduce from .
choose the constants , @xmath49 and @xmath210 , in , so that @xmath449 is the only vertex with @xmath622 and such that @xmath116 is the smallest critical value of @xmath21 greater that @xmath210 . by the markov property of the map @xmath623 for every vertex @xmath624 .
in particular let @xmath625 and let @xmath626 be the vertex at which @xmath436 intersects @xmath375 .
then by @xmath627 and by @xmath628 letting @xmath629 this becomes @xmath630 which is equal to 1 by .
thus @xmath631 . graph theoretically this is just the complete graph , @xmath338 , on @xmath308 vertices .
let us denote these vertices by @xmath633 and as in section [ sec:3 ] assign to the edge , @xmath73 , joining @xmath412 to @xmath444 , the weight @xmath634 ( as we pointed out in section [ sec:3 ] this defines an axial function on @xmath338 . )
let @xmath161 be an @xmath308-tuple of real numbers with @xmath635 and orient the edges of @xmath338 by decreeing that an edge , @xmath73 , is ascending if @xmath636 . with this orientation ,
the function mapping @xmath412 to @xmath343 is a @xmath161-compatible morse function . lets compute the thom class , @xmath637 .
if @xmath638 , we get from the computation above @xmath639 for all vertices @xmath22 .
if @xmath640 , we can regard the vertices @xmath641 , ... , @xmath642 as the vertices of a complete graph , @xmath643 , having the same axial function as above .
consider the sum @xmath644 over all ascending paths joining @xmath645 to @xmath646 , where @xmath647 .
the individual summands can be written in the form @xmath648 where by @xmath649 and @xmath650 is the product @xmath651 over all descending edge , @xmath652 , which join @xmath28 to vertices in @xmath643 .
then becomes @xmath653 however , by the expression in parentheses computes the zeroth thom class of the subgraph @xmath643 , at @xmath28 , and hence is equal to one .
thus @xmath654 where @xmath110 is the product over all the edges @xmath655 of @xmath338 , which join @xmath28 to the vertices @xmath569 , @xmath656 . | let @xmath0 be a torus of dimension @xmath1 and @xmath2 a compact hamiltonian @xmath0-manifold with @xmath3 finite . a circle , @xmath4 , in @xmath0 is _ generic _ if @xmath5 .
for such a circle the moment map associated with its action on @xmath2 is a perfect morse function .
let @xmath6 be the morse - whitney stratification of @xmath2 associated with this function , and let @xmath7 be the equivariant thom class dual to @xmath8 .
these classes form a basis of @xmath9 as a module over @xmath10 and , in particular , @xmath11 with @xmath12 . for manifolds of gkm type
we obtain a combinatorial description of these @xmath7 s and , from this description , a combinatorial formula for @xmath13 . |
when the massive black holes in galaxy centers accrete , they become visible as quasars or active galactic nuclei ( agn ) . the energy liberated by accretion episodes is thought to be a critical regulatory mechanism in galaxy evolution . but how exactly does this connection operate ? amongst the most basic unknowns is the timescale on which the agn phase operates .
indirect measurements of the typical agn phase are on the order of @xmath4 years ( e.g. * ? ? ?
* ; * ? ? ?
taking the local relic black hole mass density and comparing it to the total light emitted by quasars and assuming a reasonable radiative efficiency yields such long ( @xmath4 year ) _ total _ accretion timescales @xcite .
this does not constrain however whether the entire mass growth consists of a single accretion phase , or is broken up into shorter phases .
agn variability has been observed on short time scales all the way out to multiple decades .
the _ natural time scales _ of the agn central engine however are significantly longer than human time scales .
this leaves a major gap on the timescale domain on which we can easily study agn variability : we can see agn vary on human timescale , and we can constrain the total growth time using statistical arguments . in this paper
, we present an _
observational argument _ for why nearby agn and perhaps all agn switch on rapidly ( @xmath5 years ) and `` stay on '' for @xmath6 years before switching off . in order to reconcile this result with previous measurements of agn lifetimes
, we argue that agn _ flicker _ in a @xmath6 year cycle resulting in the total @xmath7 year lifetime .
we therefore begin to fill in agn variability between the two extremes . breaking up the accretion phases of agn into many short @xmath6 year phases
has implications beyond the fact that massive black holes assemble their mass via a large number of individual bursts .
many short phases also means that the agn central engine swings back and forth between different accretion states . during high - eddington phases
, the agn will produce predominantly radiative energy and will be visible as a classical quasar or agn . during the low - eddington phases ,
the bulk of the energy is in the form of kinetic energy . in this low - eddington phase ,
the agn is harder to detect , but may still inject ( kinetic ) energy into the host galaxy .
thus , a full accounting of the agn lifecycle and the different modes in which an agn interacts the host galaxy may be vital to a full description of agn feedback and the galaxy - black hole connection . ultimately , we will need constraints on the full power spectrum of agn from days to several gyrs to understand how black holes influence their surrounding galaxies . in this paper
, we outline the argument for a short agn lifetime and discuss some implications for our understanding of the galaxy - black hole connection and feedback .
in this section , we outline our new framework of how existing observations can constrain the duration of an individual agn phase to @xmath0 years , present these observations , and then give an order of magnitude calculation . consider the case of a central black hole in a star - forming galaxy .
when the black hole begins accreting , it become visible virtually instantaneously as a nuclear source . the accretion disk and surrounding material begin emitting at optical and ultraviolet wavelengths , and the hard x - rays will escape the nuclear region even at very high levels of obscuration . as the agn phase continues in time , photons from the central engine begin to photoionize the interstellar medium ( ism ) of the host galaxy . due to the geometry of obscuring medium , or torus ,
the photoionized region usually manifests itself as an ionization cone @xcite .
once these galaxy - scale photoionized _ narrow line regions _ are established , the host galaxy can be identified as an agn as the central engine dominates the photoionization budget of the host galaxy @xcite . *
the narrow line region ( nlr ) is therefore the part of the host galaxy interstellar medium whose emission lines are driven by agn photoionization . *
the most prominent agn line seen in optical spectra is the line , though usually a series of line ratio diagnostics is used .
higher ionization coronal lines are often also found . from this setup
, it follows that there is a significant _ time delay _ between the start of accretion in the agn central engine , and the time the host galaxy exhibits agn - like line ratios . to zeroth order ,
this time delay is on the light travel time from the nucleus to the half - light radius of the host galaxy .
naturally , the properties of the ism would be an important first order effect in this time delay as the complex physics of photoionization kicks in , but we believe that estimating the switch - on time using the light travel time across the host galaxy is the dominant effect . as the agn photoionization travels across the host galaxy , agn lines become stronger and more apparent over those driven by star formation and shocks in the galaxy , and to the host galaxy spectrum may begin to move into the `` composite '' region on emission line diagrams . ultimately the agn will photoionize the ism and the host galaxy will be classified as a regular agn .
the non - trivial light travel time of the agn radiation across the host galaxy thus constrains the _ lifetime of the agn phase _ in a statistical sense .
assuming that those agn host galaxies in a ( hard ) x - ray selected agn sample whose spectra show either pure star - formation or composite agn+star formation lines are those which just switched on , then the combination of the fraction and the light travel time across the host galaxy yields a lifetime for the agn phase : @xmath8 in order to measure the typical duration of the agn phase , we require estimates of the fraction of the agn population lacking agn lines ( the optically elusive agn or xbongs ) and the typical size of the narrow line region .
we caution that this estimate is only true in a statistical sense as it is derived from population statistics , and that there are likely many systematics in both quantities we discuss those systematics later . at a later time , once the black hole stops accreting , the ism may continue producing agn - like lines as the light echo from the past agn travels out across the galaxy . at the largest scales ,
these light echoes have been seen as the quasar light echo `` hanny s voorwerp '' near ic 2497 , and the voorwerpjes ( little voorwerps ) as well as in some quasars @xcite .
these voorwerps constrain the `` shut - down '' time scale to @xmath9 years via the geometry of their spatial extent and the recombination timescale of the narrow line emission .
we illustrate this agn phase life cycle in a schematic form in figure [ fig : schematic ] .
.literature measurements of `` optically elusive '' and xbong fractions [ cols= " < , < , < " , ] + . 6 out of 107 agn do not show a narrow line region yielding an xbong fraction of 5.6% .
this fraction is comparable to those seen in other studies @xcite .
, scaledwidth=49.0% ] if the life cycle for agn phases outlined is correct , then the switch - on phase has already been seen : x - ray - selected agn with host galaxies lacking obvious agn lines .
there is some diversity in terminology and definition of these objects in the literature .
these agn host galaxies can be described as `` optically elusive '' or `` optically dull '' agn or as xbongs ( x - ray bright , optically normal galaxies ) .
these classes can sometimes contain agn host galaxies which are red and dead and therefore do not have a sufficient cold ism to photoionize and produce strong lines , but these objects make up a minority .
such agn host galaxies have been reported repeatedly in the literature in a variety of surveys .
independent of selection and definition , the fraction of such host galaxies is consistently around the @xmath10 level as reported by various teams ; we report these values in table [ tab : samples ] . while the details of how agn are selected on emission line diagrams vary , they share generic features which go back to the original @xcite and @xcite which separate star - forming galaxies , agn and other non - stellar ionization sources using line ratios . the most common diagram and the one we use in this paper is the standard vs diagram .
star - forming galaxies are separated using an empirical line of @xcite .
the `` composite '' region where star - formation and shocks can mix extend all the way to the extreme starburst line of @xcite .
seyfert agn and liners ( low ionization nuclear emission regions ) beyond the extreme starburst line can further be separated @xcite . in general , agn photoionized galaxies are identified by having high line ratios , with a second line ratio such as separating out other processes such as shocks and evolved stellar populations . for this paper , we add an independent measurement of the `` optically elusive '' fraction .
we take the sample of hard x - ray selected agn detected by _ swift _ burst alert telescope ( bat ) in the footprint of the sloan digital sky survey in the redshift interval @xmath11 @xcite . using emission line diagnostics
, we find that 6 out of 107 agn , @xmath12 are classified as composite , meaning they are a mixture of agn and star formation .
we show the resulting classical @xcite vs diagram in figure [ fig : bpt ] .
the sdss fibers cover the nuclear region of the bat agn host galaxies so in our framework , the agn in these galaxies just switched on and has begun to photoionize the nuclear region , but has not yet overwhelmed the star formation lines .
our measurement is consistent with that of @xcite of a similar sample of _ swift _
bat agn , as well as with previous reports despite the heterogeneous nature of samples , selection techniques and definitions .
the switch - on time can be derived from the physical scale of the narrow line region ( nlr ) , which to zeroth order scales as the half - light radius of the host galaxy . what do observations
show that the nlrs of luminous agn are large and at least on scales of several kpc ( e.g. * ? ? ?
long - slit spectroscopy of obscured agn by @xcite shows that the nlr reaches the physical scale of the entire host galaxy as the agn luminosity increases and tops out , leading to a flattening of the relationship between nlr size and agn luminosity .
integral field unit studies of agn host galaxies similarly show large scale nlrs @xcite . based on these observations of nlr sizes
, we can take the half - light radius of the host galaxy of a luminous agn as a reasonable proxy for the expected nlr size . for this paper , we take the typical half - light radius of the _ swift _ bat agn host galaxy sample of @xmath13 kpc , which corresponds to @xmath14 light years . to be conservative
, we assume that the typical radius of the nlr is somewhat smaller and take a switch on time of @xmath15 years .
we now have estimates of the two quantities that go into the calculation of the agn phase lifetime .
we take the @xmath16 optically elusive fraction , and the 10,000 year switch on time scale : this yields an agn phase lifetime of : @xmath17 we stress that this estimate should be seen as an order of magnitude estimate , and that it only applies in a statistical sense .
nevertheless , we believe this typical lifetime has significant consequences for our understanding of agn variability and the role agn play in galaxy evolution . before we address these
, we discuss some of the major caveats to this estimate . in order to light up a narrow line region ,
the agn radiation needs a sufficiently dense screen of cold gas in the host galaxy to illuminate .
one of the hypotheses for the nature of xbongs is that they are gas - poor , red sequence galaxies . correspondingly , their spectra show no strong emission lines _ of any kind _ while still featuring agn x - ray emission . for the more distant agn / xbong samples
, this may be a concern , but for the _ swift _ bat sample , we can rule out this explanation for most host galaxies : as @xcite report , very few _ swift _ bat host galaxies are red and gas poor .
if such gas - poor agn hosts are a significant fraction of the total `` optical elusive '' agn fraction , this would increase the inferred agn lifetime .
a 50% contamination would double the inferred agn lifetime . in our sample of
_ swift _ bat agn , only one object features no emission lines , which implies that this is not a major bias .
another explanation for the apparent lack of agn lines in the optical spectra is that xbongs / optically elusive agn are particularly dusty and so their agn lines are not apparent in the optical spectrum .
searches for agn lines using near - infrared spectroscopy of _ swift _ bat agn host galaxies by @xcite find that most xbongs lack agn - like lines . even in the case of two objects with near - ir lines consistent with agn , star - formation remains a possibility . if a significant fraction of optically elusive agn are in fact normal agn with dust obscured nlrs , this would increase the inferred agn lifetime . as with section [ sec : gaspoor ] , a 50% contamination would double the inferred lifetime .
we assume that we can measure the switch on time using the light travel time across the host galaxy as the relevant time scale .
this is naturally a simplification as the switch on process is gradual .
more importantly , we are neglecting the well - studies physics of photoionization which certainly modulates the mapping from light travel time to appearance of narrow lines .
how much of a delay can such effects cause ? in the central parts of a galaxy , the electron density is high enough that the recombination timescale is very short . for typical values of density and temperature ,
the recombination timescale is about @xmath18 years ( see , e.g. , @xcite ) .
it is only in the outer regions of a galaxy where the density drops significantly that the recombination timescale becomes substantial ( as is the case with some voorwerps , see @xcite ) . other timescales which contribute to the delay , such as the time needed to significantly increase the ionization level , or to excite a significant fraction of ions , are expected to be shorter than the recombination timescale .
the host galaxy ism is likely already largely photoionized due to star formation , so all the agn ionization front needs to do is change the relative populations in the ions present . in any case ,
any additional delay between the agn photoionization front moving though the host galaxy and exhibiting agn lines would increase the `` optically elusive '' agn fraction artificially , and therefore further _ decrease _ the inferred lifetime . in a host galaxy with significant star formation
, the emission lines from star formation could overwhelm the agn lines , moving the spectroscopic classification out of the agn region and into the composite region .
this is a particular concern for sdss fiber spectra as the large physical footprint of the fiber can include such a signal from star formation .
@xcite showed using simulations that this dilution effect is significant for low - luminosity agn around @xmath19 but ceases to be important for brighter agn around @xmath20 .
since the _ swift _ bat agn are luminous , this dilution effect is unimportant .
furthermore , not all optically elusive samples are based on fiber spectra : both the @xcite and the @xcite studies use long - slit spectroscopy and find similar optically elusive fractions as those based on fiber spectra .
ultimately , integral field unit observations will resolve this .
agn during the switch - on phase should exhibit agn - like lines in the nucleus and lack them at larger radii despite hosting an agn sufficiently luminous ( based on x - ray observations ) to photoionize the entire galaxy . as with the caveats in section [ sec :
gaspoor ] and [ sec : dusty ] , a significant fraction of agn misclassified due to aperture effects would increase the inferred agn lifetime .
we note that while all the caveats discussed here could affect the inferred agn lifetime , the magnitude of the implied changes are on the order of a factor of @xmath21 and thus have only a moderate effect on our order - of - magnitude estimate of the agn lifetime .
we now briefly discuss several consequences of a characteristic lifetime on the order of @xmath22 years for agn phases .
we note that our argument and the resulting agn lifetime is in broad agreement with other studies , as discussed below : @xcite presented a simple model for the quasar luminosity function based on the assumption that quasar activity is driven by thermal - viscous instability in the accretion disk .
this model yields short bursts of comparable duration to our measurement driven by such instabilities and the resulting model luminosity function matches the observed data very well .
@xcite have argued that the shorter characteristic time scale of agn relative to star formation in galaxies can account for the _ apparent _ disconnect between star formation and black hole accretion .
they propose a simple model for the distribution of eddington ratios with a functional form of a power law with an exponential cutoff at high eddington ratios and allow agn to vary following this distribution . if star - forming galaxies ( _ i.e. _ those galaxies with gas ) follow this prescription for black hole accretion , the @xcite model produces a reasonable agn luminosity function and matches the observed relationship between star formation rate and black hole accretion rate .
the model does not constrain the actual agn phase lifetime , _
i.e. _ the time spent at high eddington ratio in any individual burst ; our observational argument now constrains this time .
high resolution simulations have similarly indicated short agn lifetimes : @xcite perform very high resolution hydrodynamical simulations of feeding and feedback in elliptical galaxies , resulting in short @xmath22 year agn bursts .
similarly , @xcite and @xcite find short high - eddington agn bursts in simulations of high redshift gas - rich disks with black holes . in these simulation cases ,
the enormous short time scale variability is driven by clumpy and highly variable inflow of gas in the very center of the galaxy .
all these separate approaches are converging on a picture where short bursts are the norm for black hole growth .
it remains however unclear if the burst scale is driven by the fueling mechanism ( such as the clumpiness of the material driven towards the center ) , or by instabilities in the accretion disk .
we can not resolve this issue here , but the fact that we can observationally support short individual agn phases has a number of interesting consequences : the first major consequence is that each individual black hole growth phase contributes only very little to the total mass growth of black holes .
the @xcite argument implies characteristic growth times of black holes on the order of 100 myr to 1 gyr ( see also * ? ? ?
* ) for reasonable radiative efficiencies @xmath23 .
in order to reconcile the short lifetime of @xmath22 years with a total growth time of @xmath24 , the total growth of massive black holes must consist of _ many _ ( 100 - 1000 ) such short accretion bursts .
the characteristic time scale of @xmath22 years for the agn phase must be placed in the context of both shorter- and longer - term variability as we know agn exhibit both .
on the shorter term side , there is a rich variety of observations on the hour to years time scale as many local agn are monitored and sometimes exhibit significant changes in luminosity .
this shorter - term would be superimposed on the total duration of a @xmath22 year phase , as can be seen e.g. in the simulated light curve of @xcite show in figure [ fig : novak ] . taking several @xmath22 year phases
, it seems unlikely that these would be randomly distributed .
rather , it is likely that a large number of 100 to perhaps 10,000 make up a longer growth phase of @xmath25-@xmath26 years . to put it another way ,
what we previously considered to be a @xmath27 year agn phase , perhaps triggered by a major merger , should be divided into @xmath28 short bursts . in this view , agn `` flicker '' in the same sense as a fluorescent lamp does .
the flickering behavior for agn connects the behavior of massive black holes to that of stellar mass black holes .
x - ray binaries exhibit often rapid changes in eddington ratio and accretion state which changes their appearance and their impact on their environment .
the idea that the accretion physics in x - ray binaries and agn is similar , except that agn black holes are more massive and therefore change on significantly longer time scale has been suggested a number of times . in this framework ,
the different classes of agn are closely related : black holes at a given mass rise and fall in eddington ratio , leading to appearance either as a quasar or a radio galaxy , respectively . dividing a single sustained agn phase into numerous short phases alternating between high - eddington bursts and lower - eddington troughs points to a new way in which the black hole - galaxy connection works . during each `` cycle '' , the mode via which energy is injected into the system changes : during the high - eddington peak , the agn will produce predominantly radiative energy , while during the low - eddington troughs , the change in accretion mode means that most of the energy is injected in kinetic form .
the agn flickering cycle thus alternates between radiative and kinetic modes on a short time scale compared to the dynamical time of the host galaxy .
this `` jackhammer '' effect could be important in driving agn outflows and feedback .
even if each individual cycle only injects a small amount of energy , repeating the cycle 100 - 10,000 times could add up to sufficient amounts to unbind the gas reservoir of the host galaxy as required for agn feedback . in order to approach the binding energy of a massive galaxy of @xmath29 erg
, each cycle would only have to inject @xmath30 ergs .
the short agn life cycle makes linking agn activity to outflow phenomena much more challenging .
the low - eddington phase may be difficult or impossible to detect in star - forming host galaxies and thus the galaxy may not be classified as an agn , despite the agn being active .
similarly , a kinetic outflow driven by the low - eddington phase may persist , perhaps coasting ballistically , until the next high - eddington burst .
searches for agn outflows generally target bright agn , and so might erroneously associate observed outflows with the bright phase , when in fact , the previous faint , low - eddington phase was responsible .
a short lifetime ( compared to the total growth time ) for the agn phase causes a time scale problem relative to the galaxy .
the agn central engine operate on timescales comparable only to the central dynamical time of the host galaxy , and orders of magnitude faster than the dynamical time of the whole host galaxy ( @xmath31 years for a massive galaxy ) .
the agn can switch on and off many times on the time the host galaxy at large is able to react to .
this is exacerbated by the fact that most tracers of the host galaxy star formation rate and stellar age are dependent on tracers whose time resolution is much worse than the @xmath22 time scale of the agn .
h@xmath32 is driven by photoionization of ob stars and from the lifetime of these stars alone yields a time `` smoothing kernel '' of at least @xmath33 years .
ultraviolet emission from young stars and reprocessed infrared extend even further in time as they respond to young stellar populations .
spectral features such as balmer absorption lines or the 4000 break have even worse time resolution .
this mismatch in time scales may explain why it is so difficult to link agn to galaxy evolution , and suppression of star formation in particular .
the agn lives and acts of a much shorter time scale than can be reconstructed from stellar populations .
the recent activity of the black hole in the milky way center can be revisited in light of a short agn flickering cycle .
x - ray light echoes in the galactic center show moderate - luminosity flares in the last few centuries , though these were likely too short and not sufficiently luminous to generate narrow lines across the milky way .
@xcite show that the magellanic stream near the milky way was photoionized by an agn event in the galactic center some 1 - 3 myr ago , and the duration of the burst is estimated to be 100 - 500 kyr - the same order of magnitude as the lifetime derived in this paper .
@xcite propose this galactic center outburst as the cause of the _ fermi _ bubble @xcite .
if this is the case , the flickering cycle exhibited by other agn should leave similar remnants as the _ fermi _ bubble around other galaxies as their black holes switch on and off .
we have presented an observational constraint for the typical agn phase lifetime .
the argument is based on the time lag between an agn central engine switching on and becoming visible in x - rays , and the time the agn then requires to photoionize a large fraction of the host galaxy .
based on the typical light travel time across massive galaxies , and the observed fraction of x - ray selected agn without agn - photoionized narrow lines , we estimate that the agn phase typically lasts @xmath0 years .
this short lifetime implies that black holes grow via many such short bursts and that agn therefore `` flicker '' on and off .
we discuss some consequences of this flickering behavior for agn feedback and the analogy of x - ray binaries to agn .
we thank the anonymous referee for helpful comments .
we are grateful to g. novak for sharing his simulated light curves .
we thank benny trakhtenbrot , ezequiel treister , meg urry and bill keel for useful discussions .
ks , mk and ls gratefully acknowledge support from swiss national science foundation professorship grant pp00p2_138979/1 and mk support from snsf ambizione grant pz00p2_154799/1 .
this research has made use of data from the sloan digital sky survey and the nasa swift satellite .
this research has made use of nasa s ads service . | we present an observational constraint for the typical active galactic nucleus ( agn ) phase lifetime .
the argument is based on the time lag between an agn central engine switching on and becoming visible in x - rays , and the time the agn then requires to photoionize a large fraction of the host galaxy .
based on the typical light travel time across massive galaxies , and the observed fraction of x - ray selected agn without agn - photoionized narrow lines , we estimate that the agn phase typically lasts @xmath0 years .
this lifetime is short compared to the total growth time of @xmath1 years estimated from e.g. the soltan argument and implies that black holes grow via many such short bursts and that agn therefore `` flicker '' on and off .
we discuss some consequences of this flickering behavior for agn feedback and the analogy of x - ray binaries and agn lifecycles .
4@xmath2$]/@xmath3 [ firstpage ] galaxies : active ; galaxies : seyfert ; quasars : general ; quasars : supermassive black holes |
the hot plasma in the corona of the sun and of other stars is thought to be in `` coronal equilibrium '' .
atomic excitations occur through collisions with electrons . the excited atoms decay radiatively and the emitted radiation escapes without any further interaction with the emitting plasma . as a consequence ,
the emission is optically thin , and the total flux emitted in some spectral band or in a given emission line is proportional to the emission measure @xmath15 , defined as the integral of the square of the plasma density @xmath16 over the emitting volume elements d@xmath17 through @xmath18 .
thus , observationally , the contributions of density and volume to a given observed value of @xmath15 can not be disentangled .
stellar x - ray surveys carried out with the _ einstein _ and rosat satellites have shown an enormous range of x - ray luminosity ( @xmath19 ) for stars of given spectral type ( cf . ,
@xcite , @xcite ) .
typically , one observes star to star variations in @xmath19 of up to four orders of magnitude , with the largest x - ray luminosities found among the stars with the largest rotation rates .
while one definitely finds a correlation between mean coronal temperature and x - ray luminosity ( @xcite ; @xcite ) , it is also clear that the single - most important factor contributing to the large variations in @xmath19 is the variation in emission measure .
the conclusion therefore is that active stars ( can ) have a couple of orders of magnitude higher coronal emission measure , while maintaining the same optical output as low - activity stars like our sun . the emission measure is directly linked to the structure of stellar coronae , if we assume , going along with the solar analogy , that the x - ray emitting plasma of a stellar corona is confined in magnetic loops .
the observed values of @xmath15 and @xmath19 for a given star could be accounted for either by the existence of more loops than typically visible on the solar surface , by higher density loops or by longer , more voluminous loops .
thus the question is reduced to the following : if @xmath20 for an active star , one wants to know whether this is due to @xmath21 or @xmath22 or both .
spatially resolved solar observations allow to disentangle density and volume contributions to the overall emission measure .
one finds the total x - ray output of the sun dominated at least under maximum conditions by the emission from rather small , dense loops .
stellar coronae always appear as point sources .
the only way to infer structural information in these unresolved point sources has been via eclipse studies in suitably chosen systems where one tries to constrain the emitting plasma volume from the observed light curve . another method to infer structure in spatially unresolved data are spectroscopic measurements of density .
the emissivity of plasma in coronal equilibrium in carefully selected lines does depend on density .
some lines may be present in low - density plasmas and disappear in high - density plasmas such as the forbidden lines in he - like triplets , while other lines may appear in high - density plasmas and be absent in low - density plasmas ( such as lines formed following excitations from excited levels ) . with the high - resolution spectroscopic facilities onboard _ chandra
_ it is possible to carry out such studies for a wide range of x - ray sources .
the purpose of this paper is to present and discuss some key density diagnostics available in the high - resolution grating spectra obtained with _ chandra_. we will specifically discuss the spectra obtained with the low energy transmission grating spectrometer ( letgs ) for the stars capella and procyon .
+ .[star_prop]properties of procyon and capella : mass @xmath23 , radius @xmath24 , effective temperature @xmath25 , log g and the limb darkening coefficient @xmath26 [ cols=">,>,>",options="header " , ] in all of the above analysis we assumed all triplet lines to be optically thin . in the following section we show this assumption to be consistent with our results .
let us therefore assume that the optical depth in the resonance lines is not small .
this leads to a reduction of the measured resonance line flux due to radiative scattering . at line center
the optical depth is given by the equation @xmath27 ( @xcite ) with the fractional ionization is denoted by @xmath28 , the elemental abundances by @xmath29 , the ratio of hydrogen to electron density is @xmath30 , the oscillator strength @xmath31 for all ions , the atomic number is @xmath23 , the wavelength is measured in , the temperature in k , the electron density @xmath32 in @xmath33 and the mean free path is denoted by @xmath34 . in eq .
[ tau ] we adopt a value of unity for the fractional ionization and use the solar abundances .
we further assume @xmath35 at the peak line formation and note , that @xmath36 is rather insensitive to the precise value of @xmath35 .
we can determine for each resonance line that value of @xmath37 which yields an optical depth of unity .
adopting a maximum value of @xmath32 of @xmath38 @xmath33 ( cf . tab .
[ dens_tab ] ) , we determine lower values of @xmath34 of @xmath39 cm , @xmath40 cm and @xmath41 cm , for ovii , nvi and cv , respectively . assuming a geometry most suitable for resonance scattering , we can compute the respective emission measures of @xmath42 respectively . comparing these emission measures with those derived from the measured line
fluxes @xmath43 @xmath44 with the line cooling function @xmath45 and the distance @xmath46 ( cf .
tab.[star_prop ] ) , shows the former to be larger by a few orders of magnitude for both stars .
this inconsistency shows that the assumption of a non - negligible optical depth is invalid and we conclude that optical depth effects are irrelevant for the analysis of he - like triplets in procyon and capella .
both capella and procyon are considered to be solar - like stars . in the solar context
the he - like triplets of oxygen , nitrogen and carbon have been known for a long time and in fact the he - like ion density diagnostics have been first developed to interpret these solar data .
many solar observations are available for the ovii triplet , while only very few observations have been made for the nvi and cv triplets .
the first observations of the cv triplet are reported by @xcite , who obtained an f / i - ratio of 1.9 , while @xcite found an f / i - ratio of 1.0 .
the latter authors also observed the nvi triplet , but failed to detect the intercombination line and hence deduced f / i @xmath47 1.9 .
@xcite observed the cv and nvi triplets during a flare and found a mean value of f / i @xmath48 2 with large scatter between 0.14 around this mean for nvi and a value of 0.21 ( with a scatter between 0.170.25 ) for cv . in consequence ,
little can be said about the solar nvi line ratios because of the large measurement errors .
the measured ratios for both capella and procyon are certainly consistent with the range of nvi f / i values quoted by @xcite . as far as cv is concerned
, the flare data yield very low f / i ratios indicative of high densities .
the cv f / i ratio observed for capella is higher , and that for procyon still higher than that observed for the @xcite flare we argue is due to large photospheric radiation fluxes .
we therefore conclude that the layers contributing the cv emission in capella and procyon are at lower density than those encountered in solar flares , and the same applies to the layers emitting nvi .
observations of the ovii triplet in the solar corona are reported by @xcite , @xcite , @xcite , all of which refer to the quiescent corona , while some of the observations from @xcite and @xcite refer to flares and those of @xcite refer to active regions .
in addition we considered some data points collected from various sources cited by @xcite . in fig .
[ g_r ] we plot g vs. r of these solar observations and compare these solar data with our _ chandra _ measurements for procyon and capella . for clarity we omitted individual error bars , which are typically 0.10.3 for the solar measurements .
as can be seen from fig .
[ g_r ] , most of the solar measurements yield r value between 3 and 4 . a few values are significantly lower around r @xmath48 2 , with all them referring to flares or active regions
. our _ chandra _ measurements ( 3.28 for procyon , 4.0 for capella ) thus fall into the bulk part of the solar data .
unfortunately , the solar data are by no means unambiguous .
it is not always clear which regions the data refer to ( except for the flare observations by @xcite ) , and further , most of the observations are relatively old and have not been taken with high spatial resolution ; they are comparable in some sense to our full disk observations of stellar x - ray sources .
in addition , the instrumentation used in different experiments is quite different and each experiment is affected by its own specific difficulties .
nevertheless it seems fair to state that most ovii measurements referring to quiescent conditions are close to the low density limit , thus indicating that the physical properties of the ovii emitting layers in capella and procyon should not be that different from those in the sun ; a little puzzling in this context is the large scatter of the solar g data .
in contrast , the solar flare data ( r @xmath49 2 ) is much lower than any of the other measurements , consistent with a high density plasma .
+ @xcite : quiescent ( diamond ) , flare ( asterix ) , + @xcite ( dotted diamond ) , + @xcite ( filled triangle upside down ) , + @xcite : active region ( plus symbol ) , + @xcite : flare ( cross ) .
+ encircled are our measurements of procyon ( open triangle ) and capella ( open box ) . inspection of tab .
[ dens_tab ] suggests that @xmath50 for the densities derived from cv , nvi , and ovii ; there is definitely no evidence for @xmath51 .
these findings are in contradiction to the claims made by @xcite , who argue on the basis of fexxi line ratios measured with the euve satellite that @xmath52 @xmath33 .
a recent analysis of the long wavelength part ( @xmath53 ) of the letgs spectrum of capella by @xcite , however , yielded plasma densities of @xmath54 @xmath33 at most for the plasma emitting in the fexxi lines with no evidence for any deviations from the low density limit .
of course , these values refer to those plasma layers emitting in the fexxi line , which ought to be at far higher temperature ( @xmath48 8 mk ) than those emitting in the cv , nvi , and ovii lines ( 12 mk ) .
if we assume all the observed emissions to come from the same magnetically confined structures ( loops ) , one expects the gas pressure to be more or less constant and hence the density at lower temperatures even higher .
however , plasma densities ( for capella ) of @xmath55 @xmath33 at temperatures of 2 mk are clearly inconsistent with our letgs spectra .
therefore the plasma emitting in the fexxi line is either at the low density limit or the fexxi emission comes from different structures than the emissions from he - like ions .
our temperature measurements of the line forming regions ( cf . , tab .
[ ftemp ] ) are not fully conclusive , although they suggest that the same lines of cv , nvi , and ovii respectively are formed at a somewhat higher temperature ( at most a factor of 2 ) in capella as compared to procyon . as a consequence we must have @xmath56 for the gas pressures in the respective line forming regions .
capella s peak coronal temperature is clearly much higher than procyon s . if we again assume going along with the solar analogy that for both procyon and capella ( all of ) the x - ray emission originates from magnetically confined plasma loops , the loop top temperatures in capella must be higher than in procyon .
the loop scaling laws then imply for the typical loop length scales @xmath57 for procyon and capella @xmath58 . a conservative estimate yields @xmath59 , thus @xmath14 .
the conclusion then appears inescapable that the typical sizes of the magnetic structures in procyon and capella are quite different by probably at least one order of magnitude .
next , it is instructive to compare the total mean surface fluxes emitted in the helium- and hydrogen - like ions of carbon , nitrogen and oxygen for procyon and capella . using the measured count ratios and correcting for the effects of distance , exposure times and the surface areas of the stars we find for the cv , cvi , nvi , nvii , ovii , and oviii lines values of ( capella)/(procyon ) of 0.96 , 1.37 , 1.09 , 4.9 , 1.87 , and 9.71 respectively .
therefore up to those temperatures where the cv , cvi , and nvi lines are formed , the mean surface fluxes in the two stars hardly differ at all ; pressures and temperatures are also quite similar .
we therefore conclude that the physical characteristics and global properties ( such as filling factor ) of procyon s and capella s corona at a level of @xmath60 k are very similar .
it is interesting to again perform a comparison to the sun .
@xcite quote for their sl801 rocket flight on nov 20 1969 resonance line fluxes of 4.2 10@xmath61 , 2.1 10@xmath61 , and 3.1 10@xmath61 erg/(@xmath62sec ) for ovii , nvi , and cv respectively .
correcting for the distance between earth and sun we find average surface fluxes of 200 , 100 , and 150 erg/@xmath62/sec for ovii , nvi , and cv respectively on the solar surface .
carrying out the same calculation for our _ chandra _ data , we find using the effective areas given in tab .
[ eff_ar ] mean surface fluxes in ovii of 3700 and 1980 erg/(@xmath62sec ) , in nvi of 480 and 430 erg/(@xmath62sec ) , and in cv of 830 and 860 erg/(@xmath62sec ) for capella and procyon respectively .
we therefore recover our previous finding that capella s and procyon s surface fluxes are quite similar , and determine in addition that both stars exceed typical ( ? ) solar values by one order of magnitude .
since the physical conditions of the line emitting regions are quite similar , we conclude that the coronal filling factors are larger .
evaluating now the coronal pressure for procyon we find using the densities and temperatures derived from nvi
@xmath63(nvi ) = 4.4 dyn/@xmath62 , a value also very typical for solar active regions .
from the loop scaling law ( @xcite ) @xmath64 we deduce a typical length of @xmath65 cm using @xmath66 k. this value must be considered as a lower limit to the probable length scale because of the unfortunate sensitive dependence of @xmath57 on @xmath67 through @xmath68 ; @xmath67 is likely somewhat higher than @xmath66 k as suggested by the temperatures derived from the oviii data . at any rate , however , the conclusion is , that the x - ray emission originates from low - lying loops and hence also our _ chandra _ data support the view that procyon s corona has an appearance very similar to the sun s , just like the conclusion drawn by @xcite from their euve spectra .
it then follows that the sizes of the magnetic loops in procyon s corona are similar to the loops found in the solar corona . adopting for argument s sake
a characteristic length of l@xmath69 = 1@xmath70 cm , we then find for capella @xmath71 cm , and possibly even a significant fraction of capella s radius .
this result is somewhat puzzling . on the sun , loop structures , i.e. , magnetically closed topologies ,
are occasionally found at larger heights , but they are always of low density and they never contribute significantly to the overall x - ray emission .
it will be interesting to see whether this behavior is typical for active stars in general , or whether it applies only to capella .
after all , capella may not be the prototypical active star .
its corona is not as hot as that of other stars , it does not produce flares and its radio emission is very weak .
acton l.w . , 1978 ,
apj , 255 , 1069 austin , w.e . , purcell , j.d . , tousey , r. , & widing , k.g . ,
apj , 145 , 373 blumenthal g.r . , drake g.w . & tucker w.h . ,
1972 apj , 172,205 brinkman a.c . , gunsing c.j.t . ,
kaastra j.s . , et al .
, 2000 , apj , 530 , l111 brown w.a .
, bruner m.e .
, acton l.w .
et al . , 1986 apj , 301,981 catura r.c . , acton l.w . &
johnson h.m . , 1975 ,
apjl , 196 , l47 daz - cordovs j. , claret a. , gimnez a. , 1995 , astron .
ser . , 110 , 329 doyle j.g . , 1980 ,
a&a , 87 , 183 drake j.j . , laming j.m .
, widing k.g .
, 1995 , apj , 443 , 393 dupree a.k .
, brickhouse n.s .
, doschek g.a .
et al . , 1993 ,
apjl , 418 , l41 freeman f.f . & jones b.b . , 1970 ,
15 , 288 fuhrman k. , pfeiffer m. , frank c. et al . , 1997 , astron .
, 323 , 909 gabriel a.h . &
jordan c. , 1969 , mnras , 145 , 241 hummel c.a . , armstrong j.t .
, quirrenbach a. et al . , 1994 ,
apj , 107(5 ) , 1859 hnsch m. , schmitt j.h.m.m . & voges w. , 1998a , a&a suppl .
, 127 , 251 hnsch m. , schmitt j.h.m.m . & voges w. , 1998b , a&a suppl . , 132 , 155 irwin a.w . , fletcher j.m .
, yang s.l.s .
et al . , 1992 , pasp , 104 , 489 kaastra j.s . , mewe r. , nieuwenhijzen h. , 1996 , in uv and x - ray spectroscopy of astrophysical and laboratory plasmas , k. yamashita and t. watanabe ( eds . ) , tokyo , universal academy press , inc . , p. 411
( spex ) kelch w.l .
, linsky j.l . , basri g.s .
et al . , 1978 , apj220,962 lemen j.r . , mewe r. , schrijver c. j. et al . , 1989 ,
apj , 341 , 474 linsky l.l . ,
wood b.e . ,
brown a. et al .
1998 , apj 492 , 767 mckenzie d.l .
, rugge h.r . , underwood j.h .
1978 , apj , 221 , 342 mckenzie d.l . & landecker p.b . 1982 , apj , 259 , 372 mewe r. & schrijver , j. , 1978 , astron .
astrophys . , 65 , 114 mewe r. , heise j. , gronenschild e.h.b.m .
1975 , apjl , 202 , l67 mewe r. , gronenschild e.h.b.m . , westergaard n.j .
, 1982 , apj , 260 , 233 mewe r. , gronenschild e.h.b.m .
& van den oord g.h.j .
, 1985 , a&as , 62 , 197 mewe r. , schrijver c.j . , lemen j.r .
1986 , advances in space research , 6 , 133 mewe r. , kaastra j.s .
, liedahl d.a .
, 1995 , legacy , 6 , 16 ( mekal ) mewe r. , porquet d. , raassen a.j.j .
, kaastra j.s .
, 2000a , in preparation mewe r. , raassen a.j.j . , brickhouse n.s .
et al . , 2000b submitted to a&a parkinson j.h . , 1975 ,
42 , 183 porquet & dubau 2000 , a&as , 143 , 495 pradhan a.k . , 1982 , apj , 263 , 477 pradhan a.k . , 1985 ,
apj , 288 , 824 pradhan a.k .
, norcross d.w .
& hummer d.g . , 1981 ,
apj , 246,1031 pradhan a.k . , shull j.m . , 1981 ,
apj , 249,821 predehl p. , braeuninger h. , brinkman a. et al . , 1997 ,
spie , 3113 , 172 rosner r. , tucker w.h . &
vaiana g.s . , 1978 ,
apj 220 , 643 schmitt j.h.m.m . ,
harnden f.r . ,
jr . , rosner r. , et al . , 1985 , apj , 288 , 751 schmitt j.h.m.m . , haisch b.m . & drake j.j . , 1994 ,
science , 265 , 1420 schmitt j.h.m.m .
, drake j.j . , stern r.a .
et al . , 1996a , apj , 457 , 882 schmitt j.h.m.m . , drake j.j . , haisch b.m .
et al . , 1996b , apj , 467 , 841 schmitt j.h.m.m . , 1997 , a&a , 318 , 215 schrijver , c.j . , van den oord g.h.j .
& mewe r. , 1994 , astron .
l. , 289 , l23 schrijver , c.j .
, van den oord g.h.j . ,
mewe r. et al . , 1995 , astron .
astrophys . , 302 , 438 vaiana g.s . , cassinelli j.p . , giacconi r. , 1981 , apj 244 , 163 vedder p.w .
& canizares c.r . , 1983 , apj , 270 , 666 | electron density diagnostics based on the triplets of helium - like cv , nvi , and ovii are applied to the x - ray spectra of capella and procyon measured with the low energy transmission grating spectrometer ( letgs ) on board the chandra x - ray observatory .
new theoretical models for the calculation of the line ratios between the forbidden ( f ) , intercombination ( i ) , and the resonance ( r ) lines of the helium - like triplets are used . the ( logarithmic ) electron densities ( in cgs units ) derived from the f / i ratios for capella are @xmath0 for ovii ( 2@xmath1 upper limit ) ( f / i=4.0@xmath2 ) , @xmath3 for nvi ( f / i=1.78@xmath2 ) , and @xmath4 for cv ( f / i=1.48@xmath5 ) , while for procyon we obtain @xmath6 for ovii ( f / i=3.28@xmath7 ) , @xmath8 for nvi ( f / i=1.33@xmath9 ) , and @xmath10 @xmath11 for cv ( f / i=0.48@xmath12 ) .
these densities are quite typical of densities found in the solar active regions , and also pressures and temperatures in procyon s and capella s corona at a level of @xmath13 k are quite similar .
we find no evidence for densities as high as measured in solar flares .
comparison of our capella and procyon measurements with the sun shows little difference in the physical properties of the layers producing the cv , nvi , and ovii emission . assuming the x - ray emitting plasma to be confined in magnetic loops , we obtain typical loop length scales of @xmath14 from the loop scaling laws , implying that the magnetic structures in procyon and capella are quite different .
the total mean surface fluxes emitted in the helium- and hydrogen - like ions are quite similar for capella and procyon , but exceed typical solar values by one order of magnitude .
we thus conclude that procyon s and capella s coronal filling factors are larger than corresponding solar values . |
when the characteristic size of the semiconductor device reaches the wavelength of an electron , the quantum effects become important even dominant and can not be neglected .
the accurate electromagnetic theory for the case is quantum electrodynamics ( qed ) , i.e. the second quantization for the matter and quantization for the electromagnetic field .
however , so far it is extremely difficult even impossible to employ qed to analyze the interaction between the matter and the electromagnetic field for some complex systems . the semiclassical ( or semi - quantum )
electromagnetic models are widely used in the semiconductor quantum devices .
the basic idea is that we use the maxwell s equations for the electromagnetic field while we use the schrdinger equation of the non - relativistic quantum mechanics for the matter ( see @xcite ) .
the maxwell - schrdinger coupled system ( m - s ) is written as follows : @xmath0^{2 } + q \phi(\mathbf{x},t)+v_{0 } \right\rbrace\psi(\mathbf{x},t),}\\[2 mm ] { \displaystyle \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad(\mathbf{x},t)\in\omega\times(0,t),}\\[2 mm ] { \displaystyle -\frac{\partial}{\partial t}\nabla\cdot\big(\epsilon\mathbf{a}(\mathbf{x},t)\big ) -\nabla\cdot\big(\epsilon\nabla\phi(\mathbf{x},t)\big ) = q |\psi(\mathbf{x},t)|^{2 } , \,\
, ( \mathbf{x},t)\in\omega\times(0,t ) , } \\[2 mm ] { \displaystyle \epsilon\frac{\partial ^{2}\mathbf{a}(\mathbf{x},t)}{\partial t^{2}}+\nabla\times \big({\mu}^{-1}\nabla\times \mathbf{a}(\mathbf{x},t)\big ) + \epsilon \frac{\partial ( \nabla \phi(\mathbf{x},t))}{\partial t } = \mathbf{j}_{q}(\mathbf{x},t),}\\[2 mm ] { \displaystyle \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad \qquad ( \mathbf{x},t)\in \omega\times(0,t),}\\[2 mm ] { \displaystyle \mathbf{j}_q=-\frac{\mathrm{i}q\hbar}{2m}\big(\psi^{\ast}\nabla{\psi}-\psi\nabla{\psi}^{\ast}\big)-\frac{\vert q\vert^{2}}{m}\vert\psi\vert^{2}\mathbf{a } , } \\[2 mm ] { \displaystyle \psi , \phi , \mathbf{a } \,\ , \mathrm{subject \ to \ the \ appropriate \ initial \ and\ boundary \ conditions } , } \end{array } \right.\ ] ] where @xmath1 , @xmath2 is a bounded lipschitz polyhedral convex domain , @xmath3 denotes the complex conjugate of @xmath4 , @xmath5 and @xmath6 respectively denote the electric permittivity and the magnetic permeability of the material and @xmath7 is the constant potential energy .
it is well - known that the solutions of the maxwell - schrdinger equations ( [ eq:1 - 1 ] ) are not unique .
in fact , for any function @xmath8 , if @xmath9 is a solution of ( [ eq:1 - 1 ] ) , then @xmath10 is also a solution of ( [ eq:1 - 1 ] ) .
it is often assumed that the further equations can be adjoined to the maxwell - schrdinger equations by means of a gauge transformation . in this paper
we consider the m - s system ( [ eq:1 - 1 ] ) under the temporal gauge ( also called weyl gauge ) , i.e. @xmath11 . in this paper we employ the atomic units , i.e. @xmath12 . for simplicity , we also assume that @xmath13 without loss of generality .
hence , @xmath14 and @xmath15 satisfy the following maxwell - schrdinger equations : @xmath16 { \displaystyle \frac{\partial ^{2}\mathbf{a}}{\partial t^{2}}+\nabla\times ( \nabla\times \mathbf{a } ) + \frac{\mathrm{i}}{2}\big(\psi^{*}\nabla{\psi}-\psi\nabla{\psi}^{*}\big)+\vert\psi\vert^{2}\mathbf{a}=0 , \,\ , ( \mathbf{x},t)\in \omega\times(0,t),}\\[2 mm ] \end{array } \right.\ ] ] here we omit the initial and boundary conditions for @xmath14 and @xmath15 temporarily . under the temporal gauge ,
the second equation in ( [ eq:1 - 1 ] ) involving the divergence of @xmath15 can be rewritten as @xmath17 which can be derived from ( [ eq:1 - 2 ] ) if the solutions of ( [ eq:1 - 2 ] ) are sufficiently smooth and the initial datas are consistent .
integrating with respect to @xmath18 on the both sides of ( [ eq:1 - 2 - 0 ] ) , we have @xmath19 where @xmath20 . for the purpose of theoretical analysis
, we take the gradient of ( [ eq:1 - 3 ] ) , multiply it by a parameter @xmath21 and add it to the second equation of ( [ eq:1 - 2 ] ) , to obtain @xmath22 { \displaystyle \frac{\partial ^{2}\mathbf{a}}{\partial t^{2}}+\nabla\times ( \nabla\times \mathbf{a } ) -\gamma \nabla(\nabla \cdot \mathbf{a } ) + \frac{\mathrm{i}}{2}\big(\psi^{*}\nabla{\psi}-\psi\nabla{\psi}^{*}\big)+\vert\psi\vert^{2}\mathbf{a } } \\[2 mm ] { \displaystyle \quad\,+\ , \gamma \nabla(\nabla \cdot \mathbf{a}(\mathbf{x},0 ) ) - \gamma \nabla \int_{0}^{t}\rho(\mathbf{x } , \tau ) d\tau=0 , \,\ , ( \mathbf{x},t)\in \omega\times(0,t).}\\[2 mm ] \end{array } \right.\ ] ] the parameter @xmath23 is referred to as the penalty factor .
the choice of @xmath24 depends on how much emphasis one places on the equality ( [ eq:1 - 2 - 0 ] ) . in this paper
, we keep @xmath24 fixed . to avoid the difficulty for integro - differential equations , assuming that the change of the density function @xmath25 is smooth with respect to @xmath18 for all @xmath26 , we give an approximation of @xmath27 as follows . first denoting by @xmath28 ,
we divide the time interval @xmath29 $ ] into @xmath30 subintervals @xmath31 , ( t_1 , t_2],\cdots , ( t_{m-1 } , t]$ ] . for @xmath32 $ ] ,
@xmath25 is approximated by taylor expansion and the initial conditions : @xmath33 and @xmath34
the computation of @xmath35 involves @xmath36 , the time derivative of initial wave function .
here we assume the initial conditions are consistent and so we can obtain @xmath37 from schrdinger s equation .
given an approximation of @xmath27 in @xmath38 $ ] , we can solve the coupled differential equations ( [ eq:1 - 4 ] ) in the subinterval @xmath31 $ ] and integrate the density function to obtain @xmath39 .
then for @xmath40 $ ] , @xmath27 can be calculated as follows : @xmath41 { \displaystyle\quad\quad \quad \approx \int_{0}^{t_1}\rho(\mathbf{x } , \tau ) d\tau + ( t - t_1)\rho(\mathbf{x } , t_1)+\frac{1}{2}(t - t_1)^{2 } \frac{\partial \rho}{\partial t}(\mathbf{x } , t_1 ) , \quad \forall x \in \omega}. \end{array}\ ] ] now we solve the maxwell - schrdinger equations ( [ eq:1 - 4 ] ) in the subinterval @xmath42 $ ] . repeating the above procedure , we can solve the maxwell - schrdinger equations ( [ eq:1 - 4 ] ) in the subinterval @xmath43,\cdots , ( t_{m-1 } , t]$ ] successively .
therefore , we decompose the original system in @xmath44 $ ] into m system in @xmath31 , ( t_1 , t_2],\cdots , ( t_{m-1 } , t]$ ] , respectively . for @xmath45 , j = 1 , 2,\cdots , m$ ] , the maxwell - schrdinger equations ( [ eq:1 - 4 ] ) can be rewritten as follows : @xmath46,}\\[2 mm ] { \displaystyle \frac{\partial ^{2}\mathbf{a}}{\partial t^{2}}+\nabla\times ( \nabla\times \mathbf{a})-\gamma \nabla(\nabla \cdot \mathbf{a})+\frac{\mathrm{i}}{2}\big(\psi^{*}\nabla{\psi}-\psi\nabla{\psi}^{*}\big)}\\[2 mm ] { \displaystyle \quad\quad+\,\,\vert\psi\vert^{2}\mathbf{a } = \mathbf{f}_{j}(\mathbf{x},t),\quad \,\ , ( \mathbf{x},t)\in \omega\times(t_{j-1 } , t_j],}\\[2 mm ] \end{array } \right.\ ] ] where @xmath47 is the known function .
[ rem1 - 1 ] we can get the modified maxwell - schrdinger equations ( [ eq:1 - 5 ] ) under the assumption that the change of the density function @xmath25 is smooth with respect to @xmath18 for all @xmath26 .
if the initial wave function @xmath48 is the eigenfunction of the stationary schrdinger equation and the incoming electromagnetic field is weak and can be considered as a small perturbation to the quantum system , this assumption is reasonable .
the choice of the number @xmath30 of subintervals depends on the initial wave function , the incoming electromagnetic field and @xmath49 . in this paper , we consider the following modified maxwell - schrdinger equations : @xmath50 { \displaystyle \frac{\partial ^{2}\mathbf{a}}{\partial t^{2}}+\nabla\times ( \nabla\times \mathbf{a } ) - \gamma\nabla(\nabla \cdot \mathbf{a } ) + \frac{\mathrm{i}}{2}\big(\psi^{*}\nabla{\psi}-\psi\nabla{\psi}^{*}\big)}\\[2 mm ] { \displaystyle \quad\quad+\,\,\vert\psi\vert^{2}\mathbf{a } = \mathbf{g}(\mathbf{x},t),\quad \,\ , ( \mathbf{x},t)\in \omega\times(0,t).}\\[2 mm ] \end{array } \right.\ ] ] the boundary conditions are @xmath51 and the initial conditions are @xmath52 where @xmath53 denotes the derivative of @xmath54 with respect to the time @xmath55 , @xmath56 is the outward unit normal to the boundary @xmath57 .
we assume that @xmath58 on @xmath59 .
[ rem1 - 2 ] the boundary condition @xmath60 on @xmath61 implies that the particle is confined in a whole domain @xmath62 .
the boundary condition @xmath63 on @xmath59 is referred to as the perfect conductive boundary condition .
the boundary condition @xmath64 on @xmath59 can be deduced from the boundary condition of @xmath65 and ( [ eq:1 - 3 ] ) if the initial conditions @xmath66 and @xmath67 satisfy @xmath58 on @xmath59 . as for the determination of the boundary conditions for the vector potential @xmath15 , we refer to @xcite .
[ rem1 - 3 ] the existence and uniqueness of the solution for the time - dependent maxwell - schrdinger equations ( [ eq:1 - 2 ] ) have been investigated in @xcite .
however , the results of the well - posedness of the problem were obtained only for the cauchy problem in @xmath68 , @xmath69 instead of the initial - boundary value problem . to the best of our knowledge ,
the existence and uniqueness of the solution for the maxwell - schrdinger equations in a bounded domain seem to be open .
for the modified equations ( [ eq:1 - 6])-([eq:1 - 8 ] ) , we will investigate the existence of solutions in another paper .
many authors have discussed the numerical methods for the time - dependent maxwell - schrdinger equations .
we recall some important studies about the problem .
sui and his collaborators @xcite used the finite - difference time - domain ( fdtd ) method to solve the maxwell - schrdinger equations and to simulate a simple electron tunneling problem .
pierantoni , mencarelli and rozzi @xcite applied the transmission line matrix method(tlm ) to solve the maxwell s equations and employed the fdtd method to solve the schrdinger equation , and did the simulation for a carbon nanotube between two metallic electrodes .
ahmed and li @xcite used the fdtd method for the maxwell - schrdinger system to simulate plasmonics nanodevices .
the numerical studies listed above all include a step where they extract the vector potential @xmath15 and the scalar potential @xmath70 from the electric field @xmath71 and the magnetic field @xmath72 after solving the maxwell s equations involving @xmath71 and @xmath72 .
recently , ryu @xcite employed directly the fdtd scheme to discretize the maxwell - schrdinger equations ( [ eq:1 - 1 ] ) under the lorentz gauge and to simulate a single electron in an artificial atom excited by an incoming electromagnetic field .
other related studies on this topic have been reported in @xcite and the references therein .
there are few results on the finite element method ( fem ) of the maxwell - schrdinger equations and the convergence analysis . in this paper
we will present a crank - nicolson finite element method for solving the problem ( [ eq:1 - 6])-([eq:1 - 8 ] ) , i.e. the finite element method in space and the crank - nicolson scheme in time .
then we will derive the optimal error estimates for the proposed method .
roughly speaking , compared with explicit algorithms such as the fdtd method , our method is more stable and suffers from less restriction in the time step - size since we use the crank - nicolson scheme in the time direction .
moreover , our method is more appropriate to deal with materials with discontinuous electromagnetic coefficients than the fdtd method .
our work is motivated by @xcite in which mu and huang proposed an alternating crank - nicolson method for the time - dependent ginzburg - landau equations .
the optimal error estimates were derived under the time step restrictive conditions @xmath73 for the two - dimension model and @xmath74 for the three - dimension model , where @xmath75 and @xmath76 are the spatial mesh size and the time step , respectively .
the related convergence results associated with the time - dependent ginzburg - landau equations can be also given in @xcite .
it should be emphasized that although the time - dependent ginzburg - landau model is somehow formally similar to the time - dependent maxwell - schrdinger system , there exists the essential difference between them .
the former is classified as a parabolic system and the latter is a hyperbolic system . the main key point in our work is how to avoid using the finite element inverse estimates when dealing with the nonlinear terms .
the new ideas are to derive the energy - norm error estimates for the schrdinger s equation , and to employ some tricks to eliminate the nonlinear terms both in the schrdinger s equation and in maxwell s equations , respectively .
the remainder of this paper is organized as follows . in section [ sec-2 ] ,
a crank - nicolson scheme with the galerkin finite element approximation for the modified maxwell - schrdinger equations ( [ eq:1 - 6])-([eq:1 - 8 ] ) is developed . in section [ sec-3 ] ,
the stability estimates are given .
the optimal error estimates for the numerical solution without any restriction on time step are derived in section [ sec-4 ] .
finally , the numerical testes are then carried out to confirm the theoretical results . throughout this paper , we denote by @xmath77 a generic positive constant independent of the mesh size and the time step without distinction .
in this section , we present a numerical scheme for the modified maxwell - schrdinger equations ( [ eq:1 - 6])-([eq:1 - 8 ] ) using galerkin finite element method in space and the crank - nicolson scheme in time . to start with , here and afterwards
, we assume that @xmath78 is a bounded lipschitz polygonal convex domain in @xmath79 ( or a bounded lipschitz polyhedron convex domain in @xmath80 ) .
we introduce the following notation .
let @xmath81 denote the conventional sobolev spaces of the real - valued functions . as usual , @xmath82 and @xmath83
are denoted by @xmath84 and @xmath85 respectively .
we use @xmath86 and @xmath87 with calligraphic letters for sobolev spaces of the complex - valued functions , respectively .
furthermore , let @xmath88^{d } $ ] and @xmath89^{d}$ ] with bold faced letters be sobolev spaces of the vector - valued functions with @xmath90 components ( @xmath90=2,3 ) .
@xmath91 inner - products in @xmath92 , @xmath93 and @xmath94 are denoted by @xmath95 without ambiguity .
in particular , we introduce the following subspace of @xmath96 : @xmath97 the semi - norm on @xmath98 is defined by @xmath99^{\frac{1}{2}},\ ] ] which is equivalent to the standard @xmath96-norm @xmath100 , see @xcite . to take into account the time - independence , for a time @xmath101 fixed , let @xmath102 be the bochner space defined in @xcite for @xmath103 $ ] and a banach space @xmath104 .
the weak formulation of the maxwell - schrdinger system ( [ eq:1 - 6])- ( [ eq:1 - 8 ] ) can be specified as follows : given @xmath105 , find @xmath106 such that @xmath107 , @xmath108 { \displaystyle ( \frac{\partial ^{2}\mathbf{a}}{\partial t^{2}},\mathbf{v})+(\nabla\times \mathbf{a},\nabla\times\mathbf{v } ) + \gamma(\nabla\cdot \mathbf{a},\nabla\cdot\mathbf{v})+(\frac{\mathrm{i}}{2}\big(\psi^{*}\nabla{\psi}-\psi\nabla{\psi}^{*}\big),\mathbf{v } ) } \\[2 mm ] { \displaystyle\quad \quad \quad + \ , ( |\psi|^{2}\mathbf{a},\mathbf{v})= ( \mathbf{g},\mathbf{v } ) , \qquad \qquad \qquad\qquad\qquad\forall\mathbf{v}\in\mathbf{h}^{1}_{t}(\omega ) , } \end{array } \right.\ ] ] with the initial conditions @xmath109 , @xmath110 and @xmath111 .
let m be a positive integer and let @xmath112 be the time step .
for any k=1,2,@xmath113 , we introduce the following notation : @xmath114 { \displaystyle \overline{u}^{k } = ( u^{k}+u^{k-1})/2 , \quad \widetilde{u}^{k}=(u^{k}+u^{k-2})/2,}\\[2 mm ] \end{array}\ ] ] for any given sequence @xmath115 and denote @xmath116 for any given functions @xmath117 with a banach space @xmath118 .
let @xmath119 be a regular partition of @xmath62 into triangles in @xmath79 or tetrahedrons in @xmath80 without loss of generality , where the mesh size @xmath120 .
for any @xmath121 , we denote by @xmath122 the spaces of polynomials of degree @xmath123 defined on @xmath124 .
we now define the standard lagrange finite element space @xmath125 we have the following finite element subspaces of @xmath126 , @xmath127 and @xmath128 @xmath129 we shall approximate the wave function @xmath14 and the vector potential @xmath15 by the functions in @xmath130 and @xmath131 , respectively .
let @xmath132 and @xmath133 be the conventional pointwise interpolation operators on @xmath130 and @xmath131 , respectively . for @xmath134 , @xmath135 , @xmath136 ,
standard finite element theory gives that @xcite : @xmath137 { \displaystyle \vert \mathbf{v } - { \bm \pi}_{h } \mathbf{v } \vert_{\mathbf{w}^{s}_{p } } \leq ch^{m - s}\vert \mathbf{v } \vert_{\mathbf{w}^{m}_{p } } \quad \forall \ ; \mathbf{v } \in \mathbf{w}^{m}_{p}(\omega ) . }
\end{array}\ ] ] for convenience , assume that the function @xmath15 is defined in the interval @xmath138 $ ] in terms of the time variable @xmath18 .
we can compute @xmath139 by @xmath140 which leads to an approximation to @xmath141 with second order accuracy .
a crank - nicolson galerkin finite element approximation to the maxwell - schrdinger system ( [ eq:2 - 1 ] ) is formulated as follows : @xmath142 and find @xmath143 such that for @xmath144 , @xmath145 { \displaystyle ( \partial^{2}\mathbf{a}_{h}^{k},\mathbf{v } ) + \left(\frac{\mathrm{i}}{2}\big((\psi_{h}^{k-1})^{\ast}\nabla{\psi_{h}^{k-1 } } -\psi_h^{k-1}\nabla{(\psi_{h}^{k-1})}^{\ast}\big),\mathbf{v}\right ) + ( \nabla\times \widetilde{\mathbf{a}}_{h}^{k},\nabla\times\mathbf{v})}\\[2 mm ] { \displaystyle\quad + \gamma(\nabla\cdot \widetilde{\mathbf{a}}_{h}^{k},\nabla\cdot\mathbf{v})+\big(|\psi_{h}^{k-1}|^{2}\frac{\overline{\mathbf{a}}_{h}^{k } + \overline{\mathbf{a}}_{h}^{k-1}}{2},\mathbf{v}\big)= ( \mathbf{g}^{k-1 } , \mathbf{v } ) , \quad \forall\mathbf{v}\in\mathbf{v}^{r}_{h } , } \end{array } \right.\ ] ] where @xmath146 , @xmath147 and @xmath148 have been defined in ( [ eq:2 - 2 ] ) . for convenience ,
we define the following bilinear forms : @xmath149 { \displaystyle d(\mathbf{a},\mathbf{v})=\gamma(\nabla\cdot \mathbf{a},\nabla\cdot \mathbf{v})+ ( \nabla\times\mathbf{a},\nabla\times\mathbf{v}),}\\[2 mm ] { \displaystyle f(\psi,\varphi)=\frac{\mathrm{i}}{2}(\varphi^{\ast}\nabla\psi-\psi\nabla\varphi^{\ast } ) . }
\end{array}\ ] ] then the variational forms of the modified maxwell - schrdinger equations and the discrete system can be written as follows : @xmath150 { \displaystyle ( \frac{\partial ^{2}\mathbf{a}}{\partial t^{2}},\mathbf{v } ) + d(\mathbf{a},\mathbf{v})+ ( f(\psi,\psi),\mathbf{v})+ ( |\psi|^{2}\mathbf{a},\mathbf{v})= ( \mathbf{g } , \mathbf{v } ) , \quad \forall\mathbf{v}\in\mathbf{h}^{1}_{t}(\omega ) , } \end{array } \right.\ ] ] and for @xmath151 , @xmath152 { \displaystyle ( \partial ^{2}\mathbf{a}_{h}^{k},\mathbf{v})+d(\widetilde{\mathbf{a}}_{h}^{k},\mathbf{v } ) + \left ( f(\psi_{h}^{k-1},\psi_h^{k-1}),\mathbf{v}\right)}\\[2 mm ] { \displaystyle\quad+\big ( |\psi_{h}^{k-1}|^{2}\frac{(\overline{\mathbf{a}}_{h}^{k}+\overline{\mathbf{a}}_{h}^{k-1})}{2},\mathbf{v}\big)= ( \mathbf{g}^{k-1 } , \mathbf{v } ) , \quad \forall\mathbf{v}\in\mathbf{v}^{r}_{h}. } \end{array } \right.\ ] ] note that after discretization in time and space , the maxwell equation and schrdinger equation in the discrete system ( [ eq:2 - 9 ] ) are decoupled . at each time step
, we only need to solve the two discrete linear equations alternately .
in this paper we assume that the modified maxwell - schrdinger equations ( [ eq:2 - 8 ] ) has one and only one weak solution @xmath153 and the following regularity conditions are satisfied : @xmath154 { \displaystyle\psi_{tttt } \in l^{2}(0 , t ; \mathcal{l}^{2}(\omega)),\quad \mathbf{a},\mathbf{a}_{t},\mathbf{a}_{tt } \in { l}^{\infty}(0 , t ; \mathbf{h}^{r+1}(\omega)),}\\[2 mm ] { \displaystyle\mathbf{a}_{ttt } \in { l}^{\infty}(0 , t ; \mathbf{h}^{1}(\omega)),\mathbf{a}_{tttt } \in l^{2}(0 , t ; \mathbf{l}^{2}(\omega ) ) . } \end{array}\ ] ] for the initial conditions @xmath155 and the right hand function @xmath156 , we assume that @xmath157 we now give the main convergence result in this paper as follows : [ thm2 - 1 ] suppose that @xmath1 , @xmath2 is a bounded lipschitz polyhedral convex domain .
let @xmath153 be the unique solution of the modified maxwell - schrdinger equations ( [ eq:2 - 8 ] ) , and let @xmath158 be the numerical solution of the full discrete scheme ( [ eq:2 - 9 ] ) associated with ( [ eq:2 - 8 ] ) . under the assumptions ( [ eq:2 - 10 ] ) and ( [ eq:2 - 11 ] ) , we have the following error estimates @xmath159 { \displaystyle \quad+\|\nabla\cdot(\mathbf{a}_{h}^{k}-\mathbf{a}^{k})\|_{{l}^2(\omega)}^{2 } + \|\nabla\times(\mathbf{a}_{h}^{k}-\mathbf{a}^{k})\|_{\mathbf{l}^2(\omega)}^{2}\big\ } \leq c \big\{h^{2r}+(\delta t)^{4}\big\ } , } \end{array}\ ] ] where @xmath160 , @xmath161 , and @xmath77 is a constant independent of @xmath75 , @xmath76 .
the proof of theorem [ thm2 - 1 ] will be given in section [ sec-4 ] .
in this section we derive some stability estimates for the numerical solutions of the full discrete system ( [ eq:2 - 9 ] ) , which play an important role in the error estimates in the next section . for convenience , we list the following imbedding inequalities and interpolation inequalities in sobolev spaces ( see , e.g. , @xcite and @xcite ) , and use them in the sequel : @xmath162 @xmath163 @xmath164 where @xmath165 , @xmath166 , @xmath167 and @xmath168 .
we first give the definition of the discrete energy functional of ( [ eq:2 - 9 ] ) as follows : @xmath169 { \displaystyle \quad+\frac{1}{2 } \|\partial \mathbf{a}_{h}^{k}\|_{\mathbf{l}^2}^{2}=\frac{1}{2}b(\overline{\mathbf{a}}_{h}^{k};\psi_{h}^{k},\psi_{h}^{k } ) + \frac{1}{2}\overline{d(\mathbf{a}_{h}^{k},\mathbf{a}_{h}^{k})}+v_0\|\psi_{h}^{k}\|_{\mathcal{l}^2}^{2 } + \frac{1}{2}\|\partial \mathbf{a}_{h}^{k}\|_{\mathbf{l}^2}^{2 } , } \end{array}\ ] ] where @xmath170 $ ] .
[ lem3 - 2 ] for the solution of ( [ eq:2 - 9 ] ) , for @xmath171 , we have @xmath172 where c is a constant independent of k , @xmath75 and @xmath173 . choosing @xmath174 in @xmath175 and taking the imaginary part , we can complete the proof of @xmath176 .
let us turn to the proof of @xmath177 .
it is obvious that @xmath178 = \frac{1}{2}\partial b(\overline{\mathbf{a}}_{h}^{k};{\psi}_{h}^{k},\psi_{h}^{k } ) } \\[2 mm ] { \displaystyle \quad\quad+ \frac{1}{2\delta t}\left[b(\overline{\mathbf{a}}_{h}^{k-1};{\psi}_{h}^{k-1},\psi_{h}^{k-1 } ) -b(\overline{\mathbf{a}}_{h}^{k};{\psi}_{h}^{k-1},\psi_{h}^{k-1})\right]}\\[2 mm ] { \displaystyle \quad\quad+ \frac{1}{2 \delta t}\mathrm{re}\left[b(\overline{\mathbf{a}}_{h}^{k};\psi_{h}^{k-1},\psi_{h}^{k } ) -b(\overline{\mathbf{a}}_{h}^{k};\psi_{h}^{k},\psi_{h}^{k-1})\right ] . }
\end{array}\ ] ] by a direct computation , we get @xmath179 { \displaystyle b(\mathbf{a};\psi,\varphi)-b(\hat{\mathbf{a}};\psi,\varphi)=\left((\mathbf{a } + \hat{\mathbf{a}})\psi \varphi^{*},\mathbf{a}-\hat{\mathbf{a}}\right ) + 2(f(\psi,\varphi),\mathbf{a}-\hat{\mathbf{a } } ) , } \end{array}\ ] ] and consequently @xmath180 = 0.}\ ] ] we thus have @xmath181 = -\left(|\psi_{h}^{k-1}|^{2}\frac{\overline{\mathbf{a}}_{h}^{k}+\overline{\mathbf{a}}_{h}^{k-1}}{2},\frac{\overline{\mathbf{a}}_{h}^{k } -\overline{\mathbf{a}}_{h}^{k-1}}{\delta t}\right ) } \\[2 mm ] { \displaystyle \quad+\frac{1}{2}\partial b(\overline{\mathbf{a}}_{h}^{k};{\psi}_{h}^{k},\psi_{h}^{k})-\left(f(\psi_{h}^{k-1},\psi_{h}^{k-1}),\frac{\overline{\mathbf{a}}_{h}^{k } -\overline{\mathbf{a}}_{h}^{k-1}}{\delta t}\right ) . }
\end{array}\ ] ] it is not difficult to check that @xmath182=\frac{v_{0}}{2}\partial(\psi_{h}^{k},\psi_{h}^{k}).}\ ] ] we choose @xmath183 in @xmath184 and take the real part . combining ( [ eq:3 - 11 ] ) and ( [ eq:3 - 12 ] ) gives @xmath185 { \displaystyle \quad -\left(f(\psi_{h}^{k-1},\psi_{h}^{k-1}),\frac{\overline{\mathbf{a}}_{h}^{k}-\overline{\mathbf{a}}_{h}^{k-1}}{\delta t}\right)=0 . } \end{array}\ ] ] taking @xmath186 in @xmath187 , and combining with ( [ eq:3 - 13 ] ) , we get @xmath188 { \displaystyle \quad+\partial\left(\frac{1}{4}d(\mathbf{a}_{h}^{k},\mathbf{a}_{h}^{k } ) + \frac{1}{4}d(\mathbf{a}_{h}^{k-1},\mathbf{a}_{h}^{k-1})\right)=(\mathbf{g}^{k-1},\frac{1}{2}(\partial \mathbf{a}_{h}^{k}+\partial \mathbf{a}_{h}^{k-1 } ) ) . }
\end{array}\ ] ] it follows that @xmath189 { \displaystyle \qquad \leq c\left(\vert \mathbf{g}^{k-1 } \vert_{\mathbf{l}^{2}}^{2 } + \vert \partial \mathbf{a}_{h}^{k}\vert_{\mathbf{l}^{2}}^{2 } + \vert \partial \mathbf{a}_{h}^{k-1}\vert_{\mathbf{l}^{2}}^{2}\right ) . } \end{array}\ ] ] multiply ( [ eq:3 - 13 - 0 ] ) by @xmath76 , sum @xmath190 , to discover @xmath191 now @xmath177 follows from the discrete gronwall s inequality and thus we complete the proof of lemma [ lem3 - 2 ] .
[ rem3 - 2 ] lemma [ lem3 - 2 ] shows that the numerical scheme presented in this paper for the modified maxwell - schrdinger equations ( [ eq:2 - 8 ] ) is stable in some senses .
[ thm3 - 1 ] the solution of the full discrete system ( [ eq:2 - 9 ] ) fulfills the following estimates @xmath192 { \displaystyle\quad + \|\nabla \times\mathbf{a}_{h}^{k}\|_{\mathbf{l}^2 } + \gamma\|\nabla \cdot\mathbf{a}_{h}^{k}\|_{{l}^2}\leq c , } \end{array}\ ] ] and @xmath193 where @xmath194 is a constant independent of @xmath195 , @xmath196 .
( [ eq:3 - 14 ] ) is the direct result of lemma [ lem3 - 2 ] .
next we give the proof of ( [ eq:3 - 15 ] ) . since the semi - norm in @xmath197 is equivalent to @xmath198-norm , from ( [ eq:3 - 14 ] )
we get @xmath199 then sobolev s imbedding theorem implies that @xmath200 with @xmath201 for @xmath202 and @xmath203 for @xmath204 . using young s inequality and the interpolation inequalities ( [ eq:3 - 3 ] ) ,
we further prove @xmath205 { \displaystyle \quad\leq c\|\psi^{k}_{h}\|_{\mathcal{l}^2}^{\frac{1}{2}}\|\psi^{k}_{h}\|_{\mathcal{l}^6}^{\frac{1}{2 } } \leq c\|\psi^{k}_{h}\|_{\mathcal{l}^2}^{\frac{1}{2}}\|\nabla\psi^{k}_{h}\|_{\mathbf{l}^2}^{\frac{1}{2 } } \leq c+\frac{1}{2}\|\nabla\psi^{k}_{h}\|_{\mathbf{l}^2}}. \end{array}\ ] ] hence we have @xmath206 consequently , we obtain @xmath207 combining ( [ eq:3 - 16 ] ) , ( [ eq:3 - 17 ] ) and ( [ eq:3 - 18 ] ) , we complete the proof of ( [ eq:3 - 15 ] ) . @xmath208
in this section , we will give the proof of theorem [ thm2 - 1 ] .
let @xmath209 denote the interpolation functions of @xmath210 in @xmath211 .
set @xmath212 , @xmath213 . by applying the interpolation error estimates ( [ eq:2 - 3 ] ) and
the regularity assumptions ( [ eq:2 - 10 ] ) , we have @xmath214 { \displaystyle \|i_{h}\psi\|_{\mathcal{l}^{\infty}}+\|{\bm \pi}_h\mathbf{a}\|_{\mathbf{h}^{1 } } + \|\nabla i_{h}\psi \|_{\mathbf{l}^{3}}\leq c , } \end{array}\ ] ] where @xmath194 is a constant independent of @xmath195 . for convenience , we give the following identities , which will be used frequently in the sequel . @xmath215 { \displaystyle \sum_{k=1}^{m}{(a_{k}-a_{k-1})b_{k}}=a_{m}b_{m}-a_{0}b_{0}-\sum_{k=1}^{m}{a_{k-1}(b_{k}-b_{k-1})}. } \end{array}\ ] ]
let @xmath216 , @xmath217 . by using the error estimates of the interpolation operators , we only need to estimate @xmath218 and @xmath219 . subtracting ( [ eq:2 - 8 ] ) from ( [ eq:2 - 9 ] ) ,
we get the following equations for @xmath218 and @xmath219 : @xmath220 { \displaystyle \quad+2v_0\left(\psi^{k-\frac{1}{2}}-\overline{\psi}_{h}^{k},\varphi\right)+ b(\mathbf{a}^{k-\frac{1}{2}};(\psi^{k-\frac{1}{2}}-i_{h}\overline{\psi}^{k}),\varphi)}\\[2 mm ] { \displaystyle \quad+\left(b(\mathbf{a}^{k-\frac{1}{2}};i_{h}\overline{\psi}^{k},\varphi)-b(\overline{\mathbf{a}}^{k}_{h } ; i_{h}\overline{\psi}^{k},\varphi)\right),\quad \forall \varphi\in\mathcal{v}_{h}^{r } , } \end{array}\ ] ] and @xmath221 { \displaystyle \quad + d(\mathbf{a}^{k-1}-\widetilde{{\bm \pi}_{h}\mathbf{a}^{k}},\mathbf{v } ) + \left(|\psi^{k-1}|^{2}\mathbf{a}^{k-1}-|\psi^{k-1}_{h}|^{2}\frac{\overline{\mathbf{a}}^{k}_{h}+\overline{\mathbf{a}}^{k-1}_{h}}{2 } , \;\mathbf{v}\right)}\\[2 mm ] { \displaystyle \quad+ \left(f(\psi^{k-1},\psi^{k-1})-f(\psi^{k-1}_{h},\psi^{k-1}_{h}),\;\mathbf{v}\right),\quad \forall\mathbf{v}\in\mathbf{v}^{r}_{h } , } \end{array}\ ] ] where @xmath146 , @xmath147 , @xmath222 and @xmath223 are similarly given in ( [ eq:2 - 2 ] ) .
now we briefly describe the outline of the proof of ( [ eq:2 - 12 ] ) .
first , we take @xmath224 in ( [ eq:4 - 3 ] ) and obtain the estimate of @xmath225 .
second , we choose @xmath226 in ( [ eq:4 - 3 ] ) and derive the energy - norm estimate for @xmath227 .
finally , let @xmath228 in ( [ eq:4 - 4 ] ) and acquire the estimate involving @xmath229 . combining the above three estimates
, we will complete the proof of ( [ eq:2 - 12 ] ) . to begin with ,
choosing @xmath224 , @xmath230 as the test function in ( [ eq:4 - 3 ] ) , we get @xmath231 where @xmath232 { \displaystyle i_3^{(k)}=b(\mathbf{a}^{k-\frac{1}{2}};(i_{h}\overline{\psi}^{k}-\psi^{k-\frac{1}{2}}),\overline{\theta}_{\psi}^{k } ) , \quad i_4^{(k)}=b(\overline{\mathbf{a}}^{k}_{h};i_{h}\overline{\psi}^{k},\overline{\theta}_{\psi}^{k } ) -b(\mathbf{a}^{k-\frac{1}{2}};i_{h}\overline{\psi}^{k},\overline{\theta}_{\psi}^{k } ) . }
\end{array}\ ] ] using the error estimates ( [ eq:4 - 1 ] ) for the interpolation operator @xmath233 and the regularity of @xmath14 in ( [ eq:2 - 10 ] ) , it is easy to see that @xmath234 we observe that @xmath235 { \displaystyle \quad \leq \|\nabla\psi\|_{\mathbf{l}^2}\|\nabla\varphi\|_{\mathbf{l}^2}+\|\mathbf{a}\|^{2}_{\mathbf{l}^6}\|\psi\|_{\mathcal{l}^6}\|\varphi\|_{\mathcal{l}^2 } + \|\mathbf{a}\|_{\mathbf{l}^6}\big(\|\psi\|_{\mathcal{l}^3}\|\nabla\varphi\|_{\mathbf{l}^2}}\\[2 mm ] { \displaystyle \quad+\|\nabla\psi\|_{\mathbf{l}^2}\|\varphi\|_{\mathcal{l}^3}\big)\leq c\|\nabla\psi\|_{\mathbf{l}^2 } \|\nabla\varphi\|_{\mathbf{l}^2},\quad \forall \mathbf{a}\in \mathbf{l}^6(\omega ) , \,\,\,\psi,\varphi\in\mathcal{h}_{0}^{1}(\omega ) , } \end{array}\ ] ] and @xmath236 it follows from ( [ eq:4 - 1 ] ) , ( [ eq:4 - 7 ] ) and ( [ eq:4 - 8 ] ) that @xmath237 notice that @xmath238+\big[b({\bm \pi}_{h}\overline{\mathbf{a}}^{k } ; i_{h}\overline{\psi}^{k},\overline{\theta}_{\psi}^{k})}\\[2 mm ] { \displaystyle \quad -b(\overline{\mathbf{a}}^{k } ; i_{h}\overline{\psi}^{k},\overline{\theta}_{\psi}^{k})\big]+\big[b(\overline{\mathbf{a}}^{k } ; i_{h}\overline{\psi}^{k},\overline{\theta}_{\psi}^{k } ) -b(\mathbf{a}^{k-\frac{1}{2 } } ; i_{h}\overline{\psi}^{k},\overline{\theta}_{\psi}^{k})\big]}\\[2 mm ] \end{array}\ ] ] by using ( [ eq:3 - 1])-([eq:3 - 3 ] ) and ( [ eq:3 - 10 ] ) , we prove @xmath239 { \displaystyle\qquad\quad+c\big\{h^{2r}+(\delta t)^{4}+\|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2}+\|\nabla\theta_{\psi}^{k-1}\|_{\mathbf{l}^2}^{2}\big\}. } \end{array}\ ] ] taking the imaginary part of ( [ eq:4 - 5 ] ) , we have @xmath240 { \displaystyle\qquad\quad\leq |i_1^{(k)}|+|i_2^{(k)}|+|i_3^{(k)}|+|i_4^{(k)}| \leq c \left(d ( { \theta}^{k}_{\mathbf{a}},{\theta}^{k}_{\mathbf{a } } ) + d({\theta}^{k-1}_{\mathbf{a}},{\theta}^{k-1}_{\mathbf{a}})\right)}\\[2 mm ] { \displaystyle\qquad\quad+c\big\{h^{2r}+(\delta t)^{4 } + \|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2 } + \|\nabla\theta_{\psi}^{k-1}\|_{\mathbf{l}^2}^{2}\big\ } , } \end{array}\ ] ] and therefore @xmath241 { \displaystyle \quad \leq c\big(h^{2r}+(\delta t)^{4}\big ) + c\delta t\sum_{k=1}^{m}{d({\theta}^{k}_{\mathbf{a } } , { \theta}^{k}_{\mathbf{a}})}+c\delta t\sum_{k=1}^{m } { \|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2}}. } \end{array}\ ] ] here we have used the fact @xmath242 . to proceed further ,
we take @xmath243 , @xmath230 in ( [ eq:4 - 3 ] ) , to find @xmath244 where @xmath245 { \displaystyle j_3^{(k)}=b(\mathbf{a}^{k-\frac{1}{2}};(\psi^{k-\frac{1}{2}}-i_{h}\overline{\psi}^{k}),\theta_{\psi}^{k}-\theta_{\psi}^{k-1}),}\\[2 mm ] { \displaystyle j_4^{(k)}=b(\mathbf{a}^{k-\frac{1}{2 } } ; i_{h}\overline{\psi}^{k},\theta_{\psi}^{k}-\theta_{\psi}^{k-1})-b(\overline{\mathbf{a}}^{k}_{h } ; i_{h}\overline{\psi}^{k},\theta_{\psi}^{k}-\theta_{\psi}^{k-1 } ) . }
\end{array}\ ] ] by virtue of ( [ eq:4 - 2 ] ) , we get @xmath246 { \displaystyle \quad= 2\mathrm{i}\left(\partial i_{h}\psi^{m}-(\psi_{t})^{m-\frac{1}{2}},\theta_{\psi}^{m}\right)-2\mathrm{i}\left(\partial i_{h}\psi^{1 } -(\psi_{t})^{\frac{1}{2}},\theta_{\psi}^{0}\right)}\\[2 mm ] { \displaystyle \quad-2\mathrm{i}\sum_{k=1}^{m-1}\left(\partial i_{h}\psi^{k+1 } -\partial i_{h}\psi^{k}-(\psi_{t})^{k+\frac{1}{2 } } + ( \psi_{t})^{k-\frac{1}{2}},\,\theta_{\psi}^{k}\right ) . } \end{array}\ ] ] it follows from ( [ eq:4 - 1 ] ) and ( [ eq:4 - 17 ] ) that @xmath247 to estimate the term @xmath248 , we rewrite it as @xmath249 { \displaystyle \quad-2v_0\left(\frac{1}{2}(\theta_{\psi}^{k}+\theta_{\psi}^{k-1}),\theta_{\psi}^{k}-\theta_{\psi}^{k-1}\right ) \stackrel{\mathrm{def}}{=}j_2^{(k),1}+j_2^{(k),2}. } \end{array}\ ] ] by applying a standard argument , we find that @xmath250 { \displaystyle \quad \leq c\big(h^{2r+2}+(\delta t)^{4}\big ) + c \|\theta_{\psi}^{m}\|_{\mathcal{l}^2}^{2}+ c\delta t \sum_{k=1}^{m-1}{\|\theta_{\psi}^{k}\|_{\mathcal{l}^2}^{2}}. } \end{array}\ ] ] we recall ( [ eq:3 - 10 ] ) and rewrite @xmath251 as follows : @xmath252 { \displaystyle\quad \qquad + \left(|\mathbf{a}^{k-\frac{1}{2}}|^2(\psi^{k-\frac{1}{2}}-i_{h}\overline{\psi}^{k}),\theta_{\psi}^{k}-\theta_{\psi}^{k-1}\right)}\\[2 mm ] { \displaystyle \quad \qquad+ \mathrm{i}\left(\nabla(\psi^{k-\frac{1}{2}}-i_{h}\overline{\psi}^{k})\mathbf{a}^{k-\frac{1}{2}},\;\theta_{\psi}^{k}-\theta_{\psi}^{k-1}\right)}\\[2 mm ] { \displaystyle \quad \qquad-\mathrm{i}\left((\psi^{k-\frac{1}{2}}-i_{h}\overline{\psi}^{k})\mathbf{a}^{k-\frac{1}{2}},\;\nabla \theta_{\psi}^{k } -\nabla \theta_{\psi}^{k-1}\right).}\\[2 mm ] \end{array}\ ] ] by employing ( [ eq:4 - 1 ] ) , ( [ eq:4 - 2 ] ) , the regularity assumption ( [ eq:2 - 10 ] ) and young s inequality , we can prove the following estimate of @xmath253 .
@xmath254 the proof is standard but tedious . due to space limitations
, we omit it here . in order to estimate @xmath255 ,
we rewrite @xmath256 in the following form : @xmath257}\\[2 mm ] { \displaystyle \quad\quad + \big[b(\overline{\mathbf{a}}^{k } ; i_{h}\overline{\psi}^{k},\theta_{\psi}^{k}-\theta_{\psi}^{k-1})-b({\bm \pi}_{h}\overline{\mathbf{a}}^{k } ; i_{h}\overline{\psi}^{k},\theta_{\psi}^{k}-\theta_{\psi}^{k-1})\big]}\\[2 mm ] { \displaystyle \quad\quad + \big[b({\bm \pi}_{h}\overline{\mathbf{a}}^{k } ; i_{h}\overline{\psi}^{k},\theta_{\psi}^{k}-\theta_{\psi}^{k-1})-b(\overline{\mathbf{a}}^{k}_{h } ; i_{h}\overline{\psi}^{k},\theta_{\psi}^{k}-\theta_{\psi}^{k-1})\big]}\\[2 mm ] { \displaystyle\quad\quad \stackrel{\mathrm{def}}{=}j_4^{(k),1}+j_4^{(k),2}+j_4^{(k),3}. } \end{array}\ ] ] by applying ( [ eq:3 - 15 ] ) and ( [ eq:4 - 2 ] ) , we deduce @xmath258 { \displaystyle \quad+ \frac{1}{16 } \|\nabla\theta_{\psi}^{m}\|_{\mathbf{l}^2}^{2}+c\delta t \sum_{k=1}^{m-1}{(\|\theta_{\psi}^{k}\|_{\mathcal{l}^2}^{2}+\|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2})}. } \end{array}\ ] ] in order to estimate @xmath259 , we rewrite it as follows .
@xmath260 { \displaystyle\quad\quad\quad\quad-\sum_{k=1}^{m}{\mathrm{i}\left ( i_{h}\overline{\psi}^{k}({\bm\pi}_{h}\overline{\mathbf{a}}^{k}-\overline{\mathbf{a}}^{k}_{h}),\ ; \nabla\theta_{\psi}^{k}-\nabla\theta_{\psi}^{k-1}\right)}}\\[2 mm ] { \displaystyle \quad\quad\quad\quad+\sum_{k=1}^{m}{\mathrm{i}\left(\nabla i_{h}\overline{\psi}^{k}({\bm\pi}_{h}\overline{\mathbf{a}}^{k } -\overline{\mathbf{a}}^{k}_{h}),\;\theta_{\psi}^{k}-\theta_{\psi}^{k-1}\right)}}\\[2 mm ] { \displaystyle\quad\quad\quad \stackrel{\mathrm{def}}{=}q_1+q_2+q_3 . }
\end{array}\ ] ] note that @xmath261 { \displaystyle \quad = -\left(i_{h}\overline{\psi}^{m}({\bm\pi}_{h}\overline{\mathbf{a}}^{m } + \overline{\mathbf{a}}^{m}_{h } ) \overline{\theta}_{\mathbf{a}}^{m},\;\theta_{\psi}^{m}\right ) + \left(i_{h}\overline{\psi}^{0}({\bm\pi}_{h}\overline{\mathbf{a}}^{0}+\overline{\mathbf{a}}^{0}_{h } ) \overline{\theta}_{\mathbf{a}}^{0},\;\theta_{\psi}^{0}\right)}\\[2 mm ] { \displaystyle \quad + \sum_{k=1}^{m}\left(i_{h}\overline{\psi}^{k}({\bm\pi}_{h}\overline{\mathbf{a}}^{k } + \overline{\mathbf{a}}^{k}_{h } ) \overline{\theta}_{\mathbf{a}}^{k}-i_{h}\overline{\psi}^{k-1}({\bm\pi}_{h}\overline{\mathbf{a}}^{k-1 } + \overline{\mathbf{a}}^{k-1}_{h } ) \overline{\theta}_{\mathbf{a}}^{k-1},\;\theta_{\psi}^{k-1}\right ) . } \end{array}\ ] ] by applying the young s inequality and ( [ eq:3 - 15 ] ) , we can estimate the first two terms on the right side of ( [ eq:4 - 36 ] ) as follows .
@xmath262 { \displaystyle \quad\leq \frac{1}{16}d(\overline{\theta}_{\mathbf{a}}^{m } , \overline{\theta}_{\mathbf{a}}^{m})+ c\|\theta_{\psi}^{m}\|_{\mathcal{l}^2}^{2}+ ch^{2r}. } \end{array}\ ] ] from ( [ eq:3 - 14 ] ) and ( [ eq:3 - 15 ] ) , we further deduce @xmath263 { \displaystyle \quad\leq \delta t \|i_{h}\overline{\psi}^{k}\|_{\mathcal{l}^6}\| { \bm\pi}_{h}\overline{\mathbf{a}}^{k}+\overline{\mathbf{a}}^{k}_{h}\|_{\mathbf{l}^6 } \|\frac{1}{\delta t } ( \overline{\theta}_{\mathbf{a}}^{k}-\overline{\theta}_{\mathbf{a}}^{k-1})\|_{\mathbf{l}^2}\|\theta_{\psi}^{k-1}\|_{\mathcal{l}^6}}\\[2 mm ] { \displaystyle \quad \quad+ \delta t \|\frac{i_{h}\overline{\psi}^{k}-i_{h}\overline{\psi}^{k-1}}{\delta t}\|_{\mathcal{l}^2}\| { \bm\pi}_{h}\overline{\mathbf{a}}^{k}+\overline{\mathbf{a}}^{k}_{h}\|_{\mathbf{l}^6 } \|\overline{\theta}_{\mathbf{a}}^{k-1}\|_{\mathbf{l}^6}\|\theta_{\psi}^{k-1}\|_{\mathcal{l}^6}}\\[2 mm ] { \displaystyle \quad\quad + \delta t \|i_{h}\overline{\psi}^{k-1}\|_{\mathcal{l}^6}\| \overline{\theta}_{\mathbf{a}}^{k-1}\|_{\mathbf{l}^6}\|\frac{{\bm\pi}_{h}\overline{\mathbf{a}}^{k } -{\bm\pi}_{h}\overline{\mathbf{a}}^{k-1}}{\delta t}+\frac{\overline{\mathbf{a}}_{h}^{k}-\overline{\mathbf{a}}_{h}^{k-1}}{\delta t}\|_{\mathbf{l}^2}\|\theta_{\psi}^{k-1}\|_{\mathcal{l}^6}}\\[2 mm ] { \displaystyle \quad\leq c\delta t\big\{\|\partial { \theta}_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}^{2 } + \|\partial { \theta}_{\mathbf{a}}^{k-1}\|_{\mathbf{l}^2}^{2 } + d({\theta}_{\mathbf{a}}^{k-1},{\theta}_{\mathbf{a}}^{k-1 } ) + + d({\theta}_{\mathbf{a}}^{k-2},{\theta}_{\mathbf{a}}^{k-2 } ) + \|\nabla\theta_{\psi}^{k-1}\|_{\mathbf{l}^2}^{2}\big\},}\\[2 mm ] \end{array}\ ] ] where we have used the fact : @xmath264 hence we get the following estimate : @xmath265 { \displaystyle \quad\quad\quad\leq \frac{1}{16}d(\overline{\theta}_{\mathbf{a}}^{m } , \overline{\theta}_{\mathbf{a}}^{m } ) + c\|\theta_{\psi}^{m}\|_{\mathcal{l}^2}^{2}+ c h^{2r } } \\[2 mm ] { \displaystyle \quad\quad\quad + c\delta t \sum_{k=1}^{m}\left(\|\partial { \theta}_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}^{2 } + d({\theta}_{\mathbf{a}}^{k},{\theta}_{\mathbf{a}}^{k})+ \|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2}\right ) . } \end{array}\ ] ] employing ( [ eq:4 - 1 ] ) and integrating by parts , we discover @xmath266 { \displaystyle \quad\quad + \sum_{k=1}^{m}\left(i_{h}\overline{\psi}^{k}\overline{\theta}_{\mathbf{a}}^{k}- i_{h}\overline{\psi}^{k-1}\overline{\theta}_{\mathbf{a}}^{k-1},\;\nabla\theta_{\psi}^{k-1}\right ) , } \end{array}\ ] ] by using the young s inequality , we can estimate the first three terms on the right side of ( [ eq:4 - 41 ] ) as follows : @xmath267 { \displaystyle \quad\leq \frac{1}{16}d(\overline{\theta}_{\mathbf{a}}^{m } , \overline{\theta}_{\mathbf{a}}^{m})+ c\|\theta_{\psi}^{m}\|_{\mathcal{l}^2}^{2}+c h^{2r}. } \end{array}\ ] ] using ( [ eq:4 - 1 ] ) , the last term on the right side of ( [ eq:4 - 41 ] ) can be estimated by @xmath268 { \displaystyle \quad\leq c\delta t \sum_{k=1}^{m}\left(\|\overline{\theta}_{\mathbf{a}}^{k}\|_{\mathbf{h}^{1 } } \|\nabla\theta_{\psi}^{k-1}\|_{\mathbf{l}^2}+\|\partial \overline{\theta}_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}\|\nabla\theta_{\psi}^{k-1}\|_{\mathbf{l}^2}\right)}\\[2 mm ] { \displaystyle \quad\leq c\delta t \sum_{k=0}^{m}\left(d({\theta}_{\mathbf{a}}^{k},{\theta}_{\mathbf{a}}^{k } ) + \|\partial { \theta}_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}^{2}+\|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2}\right ) . }
\end{array}\ ] ] hence we get @xmath269 { \displaystyle \quad\quad\leq \frac{1}{16}d(\overline{\theta}_{\mathbf{a}}^{m } , \overline{\theta}_{\mathbf{a}}^{m } ) + c\|\theta_{\psi}^{m}\|_{\mathcal{l}^2}^{2}+c h^{2r}}\\[2 mm ] { \displaystyle \quad \quad\quad+ c\delta t \sum_{k=0}^{m}\left(d({\theta}_{\mathbf{a}}^{k},{\theta}_{\mathbf{a}}^{k})+\|\partial { \theta}_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}^{2}+\|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2}\right ) . }
\end{array}\ ] ] reasoning as before , we can estimate @xmath270 as follows : @xmath271 { \displaystyle \quad\quad\leq \frac{1}{16}d(\overline{\theta}_{\mathbf{a}}^{m } , \overline{\theta}_{\mathbf{a}}^{m } ) + c\|\theta_{\psi}^{m}\|_{\mathcal{l}^2}^{2}+c h^{2r}}\\[2 mm ] { \displaystyle \quad\quad\quad + c\delta t \sum_{k=0}^{m}\left(d({\theta}_{\mathbf{a}}^{k},{\theta}_{\mathbf{a}}^{k})+\|\partial { \theta}_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}^{2}+\|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2}\right ) . } \end{array}\ ] ] combining ( [ eq:4 - 40 ] ) , ( [ eq:4 - 44 ] ) and ( [ eq:4 - 45 ] ) implies @xmath272 { \displaystyle \quad\quad\quad + c\delta t \sum_{k=0}^{m}\left(d({\theta}_{\mathbf{a}}^{k},{\theta}_{\mathbf{a}}^{k } ) + \|\partial { \theta}_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}^{2 } + \|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2}\right ) . }
\end{array}\ ] ] it follows from ( [ eq:4 - 30 ] ) , ( [ eq:4 - 34 ] ) and ( [ eq:4 - 46 ] ) that @xmath273 { \displaystyle \qquad+c\delta t \sum_{k=0}^{m}\big\{d({\theta}_{\mathbf{a}}^{k},{\theta}_{\mathbf{a}}^{k } ) + \|\partial { \theta}_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}^{2}+\|\theta_{\psi}^{k}\|_{\mathcal{l}^2}^{2 } + \|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2}\big\ } } \\[2 mm ] \end{array}\ ] ] now take the real part of ( [ eq:4 - 16 ] ) , and we get @xmath274 = \mathrm{re } \big(j_1^{(k)}\big)+\mathrm{re}\big(j_2^{(k)}\big)+\mathrm{re } \big(j_3^{(k)}\big)+\mathrm{re } \big(j_4^{(k)}\big ) . } \end{array}\ ] ] similarly to ( [ eq:3 - 11 ] ) , we have @xmath275= -\left(\frac{1}{2}(\overline{\mathbf{a}}_{h}^{k}+\overline{\mathbf{a}}_{h}^{k-1})|\theta_{\psi}^{k-1}|^{2},\frac{1}{2}(\partial \mathbf{a}_{h}^{k}+\partial \mathbf{a}_{h}^{k-1})\right)}\\[2 mm ] { \displaystyle\quad + \frac{1}{2}\partial b(\overline{\mathbf{a}}_{h}^{k};\theta_{\psi}^{k},\theta_{\psi}^{k})-\left(f(\theta_{\psi}^{k-1},\theta_{\psi}^{k-1}),\frac{1}{2}(\partial \mathbf{a}_{h}^{k}+\partial \mathbf{a}_{h}^{k-1})\right ) . } \end{array}\ ] ] substituting ( [ eq:4 - 49 ] ) into ( [ eq:4 - 48 ] ) and summing over @xmath276 , we get @xmath277}\\[2 mm ] { \displaystyle\quad+ \frac{1}{2}b(\overline{\mathbf{a}}_{h}^{0};\theta_{\psi}^{0},\theta_{\psi}^{0})+\delta t\sum_{k=1}^{m } \big(\frac{1}{2}(\overline{\mathbf{a}}_{h}^{k}+\overline{\mathbf{a}}_{h}^{k-1})|\theta_{\psi}^{k-1}|^{2},\frac{1}{2}(\partial \mathbf{a}_{h}^{k}+\partial \mathbf{a}_{h}^{k-1})\big)}\\[2 mm ] { \displaystyle \quad+\delta t\sum_{k=1}^{m}\big(f(\theta_{\psi}^{k-1},\theta_{\psi}^{k-1}),\frac{1}{2}(\partial \mathbf{a}_{h}^{k}+\partial \mathbf{a}_{h}^{k-1})\big ) . }
\end{array}\ ] ] combining ( [ eq:4 - 19 ] ) , ( [ eq:4 - 22 ] ) , ( [ eq:4 - 23 - 0 ] ) and ( [ eq:4 - 47 ] ) implies @xmath278}\\[2 mm ] { \displaystyle\quad \leq |\sum_{k=1}^{m}j_1^{(k)}|+|\sum_{k=1}^{m}j_2^{(k)}|+|\sum_{k=1}^{m}j_3^{(k)}|+|\sum_{k=1}^{m}j_4^{(k)}|}\\[2 mm ] { \displaystyle \leq c\left(h^{2r}+(\delta t)^{4}\right)+\frac{3}{16}d(\overline{\theta}_{\mathbf{a}}^{m } , \overline{\theta}_{\mathbf{a}}^{m } ) + \frac{1}{8 } \|\nabla\theta_{\psi}^{m}\|_{\mathbf{l}^2}^{2 } + c\|\theta_{\psi}^{m}\|_{\mathcal{l}^2}^{2}}\\[2 mm ] { \displaystyle \qquad+c\delta t \sum_{k=0}^{m}\left(d({\theta}_{\mathbf{a}}^{k},{\theta}_{\mathbf{a}}^{k } ) + \|\partial { \theta}_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}^{2}+\|\theta_{\psi}^{k}\|_{\mathcal{l}^2}^{2 } + \|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2}\right ) . } \end{array}\ ] ] by employing theorem [ thm3 - 1 ] , we discover @xmath279 setting @xmath280 and applying ( [ eq:4 - 53 ] ) and ( [ eq:4 - 53 - 0 ] ) , we obtain @xmath281 { \displaystyle \qquad + \delta t \sum_{k=1}^{m } j_5^{(k ) } + c\delta t \sum_{k=0}^{m}\left\{d({\theta}_{\mathbf{a}}^{k},{\theta}_{\mathbf{a}}^{k } ) + \|\partial { \theta}_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}^{2}+\|\theta_{\psi}^{k}\|_{\mathcal{l}^2}^{2 } + \|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2}\right\}. } \end{array}\ ] ] arguing as in the proof of theorem [ thm3 - 1 ] , we discover @xmath282 and thus @xmath283 substituting ( [ eq:4 - 57 ] ) into ( [ eq:4 - 55 ] ) , we obtain @xmath284 { \displaystyle \quad\quad\quad+c\delta t \sum_{k=0}^{m}\left\{d({\theta}_{\mathbf{a}}^{k},{\theta}_{\mathbf{a}}^{k } ) + \|\partial { \theta}_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}^{2}+\|\theta_{\psi}^{k}\|_{\mathcal{l}^2}^{2 } + \|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2}\right\}. } \end{array}\ ] ] multiplying ( [ eq:4 - 15 ] ) with @xmath285 and adding to ( [ eq:4 - 59 ] ) , we end up with @xmath286 { \displaystyle \quad\quad\quad+c\delta t \sum_{k=0}^{m}\left\{d({\theta}_{\mathbf{a}}^{k},{\theta}_{\mathbf{a}}^{k } ) + \|\partial { \theta}_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}^{2}+\|\theta_{\psi}^{k}\|_{\mathcal{l}^2}^{2 } + \|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2}\right\}. } \end{array}\ ] ] setting @xmath287 { \displaystyle k_2^{(k)}= d(\mathbf{a}^{k-1}-\widetilde{{\bm\pi}_{h}\mathbf{a}^{k}},\mathbf{v}),}\\[2 mm ] { \displaystyle k_3^{(k ) } = \big(|\psi^{k-1}|^{2}\mathbf{a}^{k-1}-|\psi^{k-1}_{h}|^{2}\frac{\overline{\mathbf{a}}^{k}_{h } + \overline{\mathbf{a}}^{k-1}_{h}}{2},\;\mathbf{v}\big),}\\[2 mm ] { \displaystyle k_4^{(k)}= \left(f(\psi^{k-1},\psi^{k-1})-f(\psi^{k-1}_{h},\psi^{k-1}_{h}),\;\mathbf{v}\right ) , } \end{array}\ ] ] we rewrite ( [ eq:4 - 4 ] ) as follows : @xmath288 we first estimate @xmath289 , @xmath290 and @xmath291 . under the regularity assumption of @xmath15 in ( [ eq:2 - 10 ] ) , we have @xmath292 by applying the regularity assumption ( [ eq:2 - 10 ] ) , the interpolation error estimates ( [ eq:4 - 1 ] ) and theorem [ thm3 - 1 ] , it is easy to deduce @xmath293 + c\sum_{k=0}^{m}\left(d({\theta}_{\mathbf{a}}^{k } , { \theta}_{\mathbf{a}}^{k})+\|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2 } + \|\mathbf{v}\|_{\mathbf{l}^2}^{2}\right).\ ] ] in order to estimate @xmath291 , we rewrite @xmath294 in the following form : @xmath295 { \displaystyle \quad + \left(f(i_{h}\psi^{k-1 } , i_{h}\psi^{k-1})-f(\psi^{k-1}_{h},\psi^{k-1}_{h}),\;\mathbf{v}\right)\stackrel{\mathrm{def}}{= } k_4^{(k),1}+k_4^{(k),2}. } \end{array}\ ] ] we observe that @xmath296 { \displaystyle = -\frac{\mathrm{i}}{2}\left(\varphi^{\ast}\nabla(\varphi-\psi)-\varphi\nabla(\varphi-\psi)^{\ast}\right ) + \frac{\mathrm{i}}{2}\left((\varphi-\psi)\nabla\psi^{\ast}-(\varphi-\psi)^{\ast}\nabla\psi\right ) . } \end{array}\ ] ] we obtain from ( [ eq:4 - 1 ] ) and ( [ eq:4 - 70 ] ) that @xmath297 similarly , from ( [ eq:4 - 1 ] ) and ( [ eq:4 - 70 ] ) , we deduce @xmath298 { \displaystyle \quad = -\frac{\mathrm{i}}{2}\left((\theta_{\psi}^{k-1})^{\ast}\nabla\theta_{\psi}^{k-1 } -\theta_{\psi}^{k-1}\nabla(\theta_{\psi}^{k-1})^{\ast},\;\mathbf{v}\right ) } \\[2 mm ] { \displaystyle \quad\quad-\frac{\mathrm{i}}{2}\left((i_{h}\psi^{k-1})^{\ast}\nabla\theta_{\psi}^{k-1 } -i_{h}\psi^{k-1}\nabla(\theta_{\psi}^{k-1})^{\ast},\;\mathbf{v}\right ) } \\[2 mm ] { \displaystyle\quad \quad + \frac{\mathrm{i}}{2}\left(\theta_{\psi}^{k-1}\nabla ( i_{h}\psi^{k-1})^{\ast}-(\theta_{\psi}^{k-1})^{\ast}\nabla i_{h}\psi^{k-1},\;\mathbf{v}\right)}\\[2 mm ] { \displaystyle \quad \leq -\left(f(\theta_{\psi}^{k-1},\theta_{\psi}^{k-1}),\mathbf{v}\right ) + c \| i_{h}\psi^{k-1}\|_{\mathcal{l}^{\infty}}\|\nabla\theta_{\psi}^{k-1}\|_{\mathbf{l}^2}\|\mathbf{v}\|_{\mathbf{l}^2}}\\[2 mm ] { \displaystyle \quad\quad+c\|\nabla i_{h}\psi^{k-1}\|_{\mathbf{l}^{3}}\|\theta_{\psi}^{k-1}\|_{\mathcal{l}^6 } \|\mathbf{v}\|_{\mathbf{l}^2}}\\[2 mm ] { \displaystyle \quad \leq -\left(f(\theta_{\psi}^{k-1},\theta_{\psi}^{k-1}),\mathbf{v}\right ) + c\left(\|\nabla\theta_{\psi}^{k-1}\|_{\mathbf{l}^2}^{2 } + \|\mathbf{v}\|_{\mathbf{l}^2}^{2}\right ) . }
\end{array}\ ] ] combining ( [ eq:4 - 69 ] ) , ( [ eq:4 - 71 ] ) and ( [ eq:4 - 72 ] ) gives @xmath299 from ( [ eq:4 - 63 ] ) , ( [ eq:4 - 68 ] ) and ( [ eq:4 - 73 ] ) , we obtain @xmath300 { \displaystyle \qquad \quad+ c\sum_{k=0}^{m}\left(d({\theta}_{\mathbf{a}}^{k } , { \theta}_{\mathbf{a}}^{k } ) + \|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2}+\|\mathbf{v}\|_{\mathbf{l}^2}^{2}\right ) . }
\end{array}\ ] ] now taking @xmath301 in ( [ eq:4 - 62 ] ) , we find @xmath302 { \displaystyle \quad = \frac{1}{2\vardelta t}\left(\|\partial \theta_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}^{2 } -\|\partial \theta_{\mathbf{a}}^{k-1}\|_{\mathbf{l}^2}^{2}\right)+\frac{1}{4\delta t}\left(d({\theta}_{\mathbf{a}}^{k } , { \theta}_{\mathbf{a}}^{k } ) -(d({\theta}_{\mathbf{a}}^{k-2 } , { \theta}_{\mathbf{a}}^{k-2})\right)}\\[2 mm ] { \displaystyle \quad = k_1^{(k)}+ k_2^{(k)}+ k_3^{(k)}+ k_4^{(k)}. } \end{array}\ ] ] note that @xmath303 thus we have @xmath304 { \displaystyle = -\sum_{k=1}^{m } j_5^{(k ) } + \sum_{k=1}^{m}\left(f(\theta_{\psi}^{k-1 } , \theta_{\psi}^{k-1}),\frac{\partial { \bm \pi}_{h}{\mathbf{a}}^{k } + \partial { \bm \pi}_{h}{\mathbf{a}}^{k-1}}{2}\right ) } \\[2 mm ] { \displaystyle \leq -\sum_{k=1}^{m } j_5^{(k ) } + c\sum_{k=1}^{m}\|\nabla\theta_{\psi}^{k-1}\|_{\mathbf{l}^{2}}^{2}}. \end{array}\ ] ] here we have used the definition of @xmath305 in ( [ eq:4 - 54 ] ) . multiplying ( [ eq:4 - 75 ] ) by @xmath173 , and using ( [ eq:4 - 74 ] ) and ( [ eq:4 - 75 - 0 ] ) ,
we obtain @xmath306 { \displaystyle \quad \leq
c\left\{h^{2r}+(\delta t)^{4 } \right\ } + c\delta t\sum_{k=0}^{m}\left(d({\theta}_{\mathbf{a}}^{k } , { \theta}_{\mathbf{a}}^{k})+\|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2 } + \|\partial \theta_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}^{2 } \right ) } \\[2 mm ] { \displaystyle \quad+ \delta t \sum_{k=1}^{m } k_2^{(k ) } -\delta t\sum_{k=1}^{m } j_5^{(k)}. } \end{array}\ ] ] since @xmath307 , by applying ( [ eq:4 - 2 ] ) and the young s inequality , we get @xmath308 { \displaystyle \leq c\left\{h^{2r } + ( \delta t)^{4}\right\ } + \frac{1}{32 } d\left(\theta_{\mathbf{a}}^{m},\theta_{\mathbf{a}}^{m}\right ) + \frac{1}{32 } d\left(\theta_{\mathbf{a}}^{m-1},\theta_{\mathbf{a}}^{m-1}\right ) + c\delta t \sum_{k=0}^{m}d\left(\theta_{\mathbf{a}}^{k},\theta_{\mathbf{a}}^{k}\right)}\\[2 mm ] \end{array}\ ] ] combining ( [ eq:4 - 60 ] ) , ( [ eq:4 - 76 ] ) and ( [ eq:4 - 77 ] ) implies @xmath309 { \displaystyle \quad\leq c\left\{h^{2r}+(\delta t)^{4}\right\ } + c\delta t \sum_{k=0}^{m}\left\{d({\theta}_{\mathbf{a}}^{k},{\theta}_{\mathbf{a}}^{k } ) + \|\partial { \theta}_{\mathbf{a}}^{k}\|_{\mathbf{l}^2}^{2}+\|\theta_{\psi}^{k}\|_{\mathcal{l}^2}^{2 } + \|\nabla\theta_{\psi}^{k}\|_{\mathbf{l}^2}^{2}\right\}. } \end{array}\ ] ] by applying the discrete gronwall s inequality , we have @xmath310 combine ( [ eq:4 - 82 ] ) with the interpolation error estimates ( [ eq:4 - 1 ] ) and we can complete the proof of theorem [ thm2 - 1 ] .
to validate the developed algorithm and to confirm the theoretical analysis reported in this paper , we present numerical simulations for the following case studies . [ exam6 - 1 ] we consider the maxwell - schrdinger system ( [ eq:1 - 2 ] ) , where the initial - boundary conditions are as follows : @xmath311 { \displaystyle \psi(\mathbf{x},0 ) = 2\sqrt{2}\sin(\pi x_1)\sin(\pi x_2)\sin(\pi x_3 ) , \quad \mathbf{a}_{t}(\mathbf{x},0)=\mathbf{a}_1(\mathbf{x})=0 , } \\[2 mm ] { \mathbf{a}(\mathbf{x},0)=\mathbf{a}_0(\mathbf{x } ) = \big(10x_1 x_2 x_3(1-x_2)(1-x_3 ) , 10x_1 x_2 x_3(1-x_1)(1-x_3 ) , } \\[2 mm ] { \displaystyle\qquad \qquad \qquad \qquad 10x_1 x_2 x_3(1-x_1)(1-x_2)\big ) . }
\end{array}\ ] ] here we take @xmath312 , @xmath313 , @xmath314 , and the time step @xmath315 . note that the initial wave function @xmath48 is the eigenfunction of the stationary schrdinger s equation .
the numerical results are displayed in fig .
5.1 . :
( a ) the evolution of the density function @xmath316 on the line @xmath317 at time @xmath318 ; ( b ) the evolution of @xmath319 , @xmath320 and @xmath321 , where @xmath322 , @xmath323 and @xmath324 . , title="fig:",width=226,height=226 ] : ( a ) the evolution of the density function @xmath316 on the line @xmath317 at time @xmath318 ; ( b ) the evolution of @xmath319 , @xmath320 and @xmath321 , where @xmath322 , @xmath323 and @xmath324 . , title="fig:",width=226,height=226 ]
[ fig:6 - 1 ] [ rem6 - 1 ] the numerical results illustrated in fig .
5.1 clearly show that the change of @xmath25 is smooth with respect to @xmath18 and the assumption on which the modified maxwell - schrdinger equations are based is valid in this case .
[ exam6 - 2 ] we consider the modified maxwell
schrdinger s equations as follows : @xmath325 { \displaystyle \frac{\partial ^{2}\mathbf{a}}{\partial t^{2}}+\nabla\times ( \nabla\times \mathbf{a } ) - \gamma\nabla(\nabla \cdot \mathbf{a } ) + \frac{\mathrm{i}}{2}\big(\psi^{*}\nabla{\psi}-\psi\nabla{\psi}^{*}\big)}\\[2 mm ] { \displaystyle \quad\quad+\,\,\vert\psi\vert^{2}\mathbf{a } = \mathbf{g}(\mathbf{x},t),\quad \,\ , ( \mathbf{x},t)\in \omega\times(0,t ) , } \end{array } \right.\ ] ] where the initial - boundary conditions are given in ( [ eq:1 - 7])-([eq:1 - 8 ] ) .
we take @xmath326 , @xmath327 , @xmath328 and @xmath329 . the exact solution @xmath153 of ( [ eq:6 - 11 ] ) is given by @xmath330 { \displaystyle \qquad\qquad\qquad+ 5.0e^{\mathrm{i}\pi t } \sin(2\pi x_1)\sin(2\pi x_2)\sin(2\pi x_3 ) , } \end{array}\ ] ] @xmath331 { \displaystyle \quad \sin(2\pi x_1)\sin(2\pi x_2)\cos(2\pi x_3)\big ) + \cos(\pi t)\big(\cos(\pi x_1)\sin(\pi x_2)\sin(\pi x_3),}\\[2 mm ] { \displaystyle \quad \sin(\pi x_1)\cos(\pi x_2)\sin(\pi x_3 ) , \sin(\pi x_1)\sin(\pi x_2)\cos(\pi x_3)\big ) . }
\end{array}\ ] ] the functions @xmath332 and @xmath333 in ( [ eq:6 - 11 ] ) are chosen correspondingly to the exact solution @xmath153 .
a uniform tetrahedral partition is generated with @xmath334 nodes in each direction and @xmath335 elements in total .
we solve the system([eq:6 - 11 ] ) by the proposed crank - nicolson galerkin finite element scheme ( [ eq:2 - 9 ] ) with linear elements and quadratic elements , respectively . to confirm our error analysis
, we take @xmath336 for the linear element method and @xmath337 for the quadratic element method respectively .
numerical results for the linear element method and the quadratic element method at time @xmath338 are listed in tables [ table6 - 1 ] and [ table6 - 2 ] , respectively .
.@xmath339 error of linear fem with @xmath340 and @xmath341 . [ cols="^,^,^,^,^",options="header " , ] [ rem6 - 2 ] numerical results in tables [ table6 - 1 ] and [ table6 - 2 ] are in good agreement with the theoretical analysis , see theorem [ thm2 - 1 ] .
[ exam6 - 3 ] we consider the following modified maxwell
schrdinger s equations @xmath342 { \displaystyle \frac{\partial ^{2}\mathbf{a}}{\partial t^{2}}+\nabla\times ( \nabla\times \mathbf{a } ) - \gamma\nabla(\nabla \cdot \mathbf{a } ) + \frac{\mathrm{i}}{2}\big(\psi^{*}\nabla{\psi}-\psi\nabla{\psi}^{*}\big)}\\[2 mm ] { \displaystyle \quad\quad+\,\,\vert\psi\vert^{2}\mathbf{a } = \mathbf{g}(\mathbf{x},t),\quad \,\ , ( \mathbf{x},t)\in \omega\times(0,t).}\\[2 mm ] { \displaystyle \psi(\mathbf{x},t)=0,\quad \mathbf{a}(\mathbf{x},t)\times\mathbf{n}=0 , \quad \nabla \cdot \mathbf{a}(\mathbf{x},t ) = 0 , \quad ( \mathbf{x},t)\in \partial \omega\times(0,t),}\\[2 mm ] { \displaystyle \psi(\mathbf{x},0 ) = \psi_0(\mathbf{x}),\quad\mathbf{a}(\mathbf{x},0)=\mathbf{a}_{0}(\mathbf{x}),\quad \mathbf{a}_{t}(\mathbf{x},0)=\mathbf{a}_{1}(\mathbf{x } ) , } \end{array } \right.\ ] ] with @xmath343 { \displaystyle \mathbf{g}(\mathbf{x } , t ) = \left(10\sin(1.5{\pi}^{2 } t ) , 10\sin(1.5{\pi}^{2 } t ) , 10\cos(1.5{\pi}^{2 } t)\right ) . } \end{array}\ ] ] in this example we take @xmath326 , @xmath344 , @xmath345 , @xmath346 . using the mesh in example [ exam6 - 2 ] with m = 50 , we solve the system ( [ eq:6 - 12 ] ) by the proposed crank - nicolson galerkin finite element scheme ( [ eq:2 - 9 ] ) with linear elements .
the time step @xmath315 . in fig .
5.2 we display the numerical results of @xmath347 on the line @xmath317 and on the intersection @xmath348 at time @xmath349 . in fig .
5.3 we plot the numerical results of @xmath347 on the intersections @xmath350 and @xmath351 at time @xmath352 , respectively . :
numerical results of the density function @xmath347 on the line @xmath317 and the contour plots of @xmath347 on the intersection @xmath348 at time @xmath349 .
( a ) @xmath347 on the line @xmath317 at time @xmath349 ; ( b ) the contour plot at time @xmath353 ; ( c ) the contour plot at time @xmath354 ; ( d ) the contour plot at time @xmath355 ; ( e ) the contour plot at time @xmath356 ; ( f ) the contour plot at time @xmath352 ; , title="fig:",width=226,height=226 ] : numerical results of the density function @xmath347 on the line @xmath317 and the contour plots of @xmath347 on the intersection @xmath348 at time @xmath349 .
( a ) @xmath347 on the line @xmath317 at time @xmath349 ; ( b ) the contour plot at time @xmath353 ; ( c ) the contour plot at time @xmath354 ; ( d ) the contour plot at time @xmath355 ; ( e ) the contour plot at time @xmath356 ; ( f ) the contour plot at time @xmath352 ; , title="fig:",width=226,height=226 ] : numerical results of the density function @xmath347 on the line @xmath317 and the contour plots of @xmath347 on the intersection @xmath348 at time @xmath349 .
( a ) @xmath347 on the line @xmath317 at time @xmath349 ; ( b ) the contour plot at time @xmath353 ; ( c ) the contour plot at time @xmath354 ; ( d ) the contour plot at time @xmath355 ; ( e ) the contour plot at time @xmath356 ; ( f ) the contour plot at time @xmath352 ; , title="fig:",width=226,height=226 ] : numerical results of the density function @xmath347 on the line @xmath317 and the contour plots of @xmath347 on the intersection @xmath348 at time @xmath349 .
( a ) @xmath347 on the line @xmath317 at time @xmath349 ; ( b ) the contour plot at time @xmath353 ; ( c ) the contour plot at time @xmath354 ; ( d ) the contour plot at time @xmath355 ; ( e ) the contour plot at time @xmath356 ; ( f ) the contour plot at time @xmath352 ; , title="fig:",width=226,height=226 ] : numerical results of the density function @xmath347 on the line @xmath317 and the contour plots of @xmath347 on the intersection @xmath348 at time @xmath349 .
( a ) @xmath347 on the line @xmath317 at time @xmath349 ; ( b ) the contour plot at time @xmath353 ; ( c ) the contour plot at time @xmath354 ; ( d ) the contour plot at time @xmath355 ; ( e ) the contour plot at time @xmath356 ; ( f ) the contour plot at time @xmath352 ; , title="fig:",width=226,height=226 ] : numerical results of the density function @xmath347 on the line @xmath317 and the contour plots of @xmath347 on the intersection @xmath348 at time @xmath349 .
( a ) @xmath347 on the line @xmath317 at time @xmath349 ; ( b ) the contour plot at time @xmath353 ; ( c ) the contour plot at time @xmath354 ; ( d ) the contour plot at time @xmath355 ; ( e ) the contour plot at time @xmath356 ; ( f ) the contour plot at time @xmath352 ; , title="fig:",width=226,height=226 ] : numerical results of the density function @xmath347 on the intersection @xmath357 and @xmath358 at time @xmath359 .
( a ) @xmath347 on the intersection @xmath360 ; ( b ) @xmath347 on the intersection @xmath361 ; ( c ) @xmath347 on the intersection @xmath358 ; , title="fig:",width=226,height=226 ] : numerical results of the density function @xmath347 on the intersection @xmath357 and @xmath358 at time @xmath359 .
( a ) @xmath347 on the intersection @xmath360 ; ( b ) @xmath347 on the intersection @xmath361 ; ( c ) @xmath347 on the intersection @xmath358 ; , title="fig:",width=226,height=226 ] : numerical results of the density function @xmath347 on the intersection @xmath357 and @xmath358 at time @xmath359 .
( a ) @xmath347 on the intersection @xmath360 ; ( b ) @xmath347 on the intersection @xmath361 ; ( c ) @xmath347 on the intersection @xmath358 ; , title="fig:",width=226,height=226 ] althouth the maxwell - schrdinger system and the time - dependent ginzburg - landau equations are somehow formally similar , they describe the different physical phenomenons .
the time - dependent ginzburg - landau equations describe the vortex dynamics of superconductor @xcite while the maxwell - schrdinger equations describe the wave packet dynamics of an electron . as can be seen in fig .
5.2 - 5.3 , the wave packet of the electron is located at the center of the computational domain at first and the external and its self - induced electromagnetic fields cause the motion of the wave packet .
unlike the time - dependent ginzburg - landau equations , no stable state is observed in our computation .
we have presented the optimal @xmath339 error estimates of a crank - nicolson galerkin finite element method for the modified maxwell - schrdinger equations , which are derived from the original equations under some assumptions .
the techniques used in this paper may also be applied to other nonlinear pdes , such as the ginzburg - landau equations . the original maxwell - schrdinger system is challenging and difficult to perform numerical computation and theoretical analysis .
our work can serve as an elementary attempt for the numerical analysis of this system .
we will study the original system in a further work using the mixed finite element method .
, _ coupled analysis of maxwell - schrdinger equations by using the length gauge : harmonic model of a nanoplate subjected to a 2d electromagnetic field _ , intern .
j. of numer . model . : electronic networks , devices and fields ( 2013 ) , 26(6 ) : pp .
533544 . ,
_ a new 3-d transmission line matrix scheme for the combined schrdinger - maxwell problem in the electronic / electromagnetic characterization of nanodevices _ , ieee transactions on microwave theory and techniques ( 2008 ) , 56(3 ) : pp . 654. | in this paper we consider the initial - boundary value problem for the time - dependent maxwell - schrdinger equations , which arises in the interaction between the matter and the electromagnetic field for the semiconductor quantum devices . a crank - nicolson finite element method for solving the problem
is presented .
the optimal energy - norm error estimates for the numerical algorithm without any time - step restrictions are derived .
numerical tests are then carried out to confirm the theoretical results .
time - dependent maxwell - schrdinger equations , finite element method , crank - nicolson scheme , optimal error estimate . 65n30 , 65n55 , 65f10 , 65y05 |
existence and properties of closed geodesics on riemannian manifolds have been subject of intense research since poincar s work @xcite from the beginning of the 20th century . a prominent result in the field
is a theorem by gromoll and meyer that guarentees the existence of infinitely many closed geodesics on compact riemannian manifolds on some assumption on the cohomology ring @xcite .
this assumption is satisfied by a large class of manifolds but not by spheres .
generalizing ideas of birkhoff @xcite , franks @xcite and bangert @xcite together proved the existence of infinitely many closed geodesics on every riemannian @xmath0-sphere . in @xcite guruprasad and haefliger
generalized the result by gromoll and meyer to the setting of riemannian orbifolds . in the present paper
we generalize the result by bangert and franks in the following way .
[ thm : inf_geo ] on every compact riemannian @xmath0-orbifold there exist infinitely many geometrically distinct closed geodesics of positive length . for spindle orbifolds
this statement was before only known in the rotational symmetric case @xcite .
in @xcite the existence of a single geodesic in the regular part of any @xmath0-orbifold with isolated singularities is claimed .
the alleged geodesic in the regular part is obtained as a limit of locally length minimizing curves contained in the complement of shrinking @xmath1-neighborhoods of the singular points .
note , however , that there are examples in which such curves have a limit that is not contained in the regular part .
for instance , on a plane with two singular points of cone angle @xmath2 which is otherwise flat , curves of minimal length that enclose both @xmath1-neighborhoods converge to a straight segment between the singular points . to prove our result we first reduce its statement to the case of simply connected spindle orbifolds ( see section [ sec : pre ] ) . using the curve - shortening flow we are able to prove the existence of an embedded geodesic in the regular part of any simply connected spindle orbifold and to apply ideas from bangert s proof and frank s result in this case .
our proof relies on the observation that embedded loops in the regular part that evolve under the curve - shortening flow either stay in the regular part forever , or collapse into a singular point in finite time ( see section [ sec : curve_shortening ] ) .
the possibility of a limit curve being not contained in the regular part is then excluded by topological arguments ( see proposition [ prp : simple_geodesic ] ) .
recall that a _ length space _ is a metric space in which the distance of any two points can be realized as the infimum of the lengths of all rectifiable paths connecting these points , cf .
a riemannian orbifold can be defined as follows .
an _ @xmath3-dimensional riemannian orbifold _
@xmath4 is a length space such that for each point @xmath5 , there exists a neighborhood @xmath6 of @xmath7 in @xmath4 , an @xmath3-dimensional riemannian manifold @xmath8 and a finite group @xmath9 acting by isometries on @xmath8 such that @xmath6 and @xmath10 are isometric . behind this definition
lies the fact that an effective , isometric action of a finite group on a simply connected riemannian manifold can be recovered from the corresponding metric quotient @xcite .
in particular , the underlying topological space of a riemannian orbifold in this sense admits a smooth orbifold structure and a compatible riemannian structure in the usual sense ( cf .
@xcite ) that in turn induces the metric . for a point @xmath7 on a riemannian orbifold ,
the isotropy group of a preimage of @xmath7 in a riemannian manifold chart is uniquely determined up to conjugation .
its conjugacy class in @xmath11 is denoted as @xmath12 and is called the _ local group _ of @xmath4 at @xmath7 .
the points with trivial local group form the _ regular part _ of @xmath4 , which is a riemannian manifold .
all other points are called _
singular_. we will be particularly concerned with @xmath0-orbifolds all of whose singular points are isolated . in this case all local groups are cyclic and we refer to their orders as the _ orders of the singular points_. we are interested in ( orbifold ) geodesics defined in the following way .
[ dfn : orb_geo ] an ( orbifold ) _ geodesic _ on a riemannian orbifold is a continuous path that can locally be lifted to a geodesic in a riemannian manifold chart .
a closed ( orbifold ) geodesic is a continuous loop that is a ( orbifold ) geodesic on each subinterval . in the following , by a ( closed )
geodesic we always mean a ( closed ) orbifold geodesic .
a geodesic that hits an isolated singularity is not locally length minimizing . on a @xmath0-orbifold such
a geodesic is either reflected or goes straight through the singular point depending on whether the order of the ( local group of the ) singular point is even or odd .
we say that two geodesics are _ geometrically distinct _ if their geometric trajectories differ .
given a closed geodesic @xmath13 the iterations @xmath14 , @xmath15 , form a whole _ tower _
of geometrically equivalent closed geodesics . in the following , by infinitely many closed geodesics we always mean infinitely many geometrically distinct closed geodesics of positive length .
we need the following concept .
a _ covering orbifold _ or _ orbi - cover _ of a riemannian orbifold @xmath4 is a riemannian orbifold @xmath16 together with a surjective map @xmath17 such that each point @xmath18 has a neighborhood @xmath6 isometric to some @xmath10 for which each connected component @xmath19 of @xmath20 is isometric to @xmath21 for some subgroup @xmath22 such that the isometries respect the projections .
-spindle orbifold @xmath23 , i.e. a @xmath0-orbifold with at most two isolated singularities of order @xmath24 and @xmath25 ( with @xmath24 or @xmath25 perhaps being @xmath26 ) .
note that @xmath27-spindle orbifolds are also known as teardrops .
the orbifolds in the picture are bad if and only if @xmath28 and simply connected as orbifolds if and only if @xmath24 and @xmath25 are coprime.,scaledwidth=25.0% ] an orbifold is called _ simply connected _ , if it does not admit a nontrival orbi - cover .
an orbifold is called _ good _ ( or developable ) if it is covered by a manifold .
otherwise it is called _ bad _ @xcite .
the only bad @xmath0-orbifolds are depicted in figure [ spindle ] , cf .
in fact , every compact good @xmath0-orbifold is _ very good _ meaning that it is finitely covered by a ( necessarily compact ) manifold ( * ? ? ? * thm .
2.5 ) . clearly , if an orbifold is finitely covered by an orbifold with infinitely many closed geodesics , then it has itself infinitely many closed geodesics .
since all riemannian surfaces are known to have infinitely many geodesics , in view of proving our main result it suffices to treat simply connected _ spindle orbifolds _
, i.e. spindle orbifolds @xmath23 with @xmath24 and @xmath25 coprime ( see figure [ spindle ] ) .
we would like to apply morse theory and homological methods to find closed geodesics on orbifolds .
to this end a notion of a loop space is needed .
such a notion is defined in @xcite .
to any compact riemannian orbifold a free loop space @xmath29 is associated and endowed with a natural structure of a complete riemannian hilbert orbifold .
we sketch this construction in the appendix , section [ app : loop_space ] . here
we give an alternative description of @xmath29 in the case in which @xmath4 has only isolated singularities .
let @xmath30 be a loop on @xmath4 . in the following we always assume that such a loop @xmath31 is of class @xmath32 , i.e. that it locally lifts to @xmath32 curves on manifold charts .
[ dfn : develop ] a _ development of @xmath30 _ is a loop @xmath33 on a riemannian manifold @xmath8 together with a map @xmath34 which is locally an orbi - covering and which projects @xmath33 to @xmath30 .
the development is called _ geodesic _ if @xmath33 is a geodesic on @xmath8 .
every loop on @xmath4 can be locally lifted to riemannian manifold charts .
a development of a loop @xmath30 can be obtained by gluing together the riemannian manifold charts that support the local lifts . two developments @xmath35 and @xmath36
are said to be equivalent if there exist neighborhoods @xmath37 of @xmath38 in @xmath39 and @xmath40 of @xmath41 in @xmath42 and an isometry @xmath43 that maps @xmath38 to @xmath41 .
[ dfn : orbi_loop ] an _ orbifold loop _ is a loop @xmath30 on @xmath4 together with an equivalence class of developments of @xmath30 .
the orbifold loop is called _ geodesic _ if the developments are geodesic .
this notion is compatible with the notion of an orbifold geodesic .
every geodesic orbifold loop projects to an orbifold geodesic in the sense of definition [ dfn : orb_geo ] and every orbifold geodesic admits a unique equivalence class of geodesic developments . as a set the free loop space
@xmath29 is the collection of all orbifold loops . to describe its metric stucture we proceed as follows
let @xmath44 be a development defining an orbifold loop @xmath30 .
the free loop space @xmath45 has a natural structure of a riemannian hilbert manifold @xcite and we have @xmath46 .
a loop @xmath47 can be regarded as an orbifold loop represented by the development @xmath48 .
if @xmath30 is not a constant loop at a singular point of @xmath4 , then we can choose a neighborhood @xmath49 of @xmath33 in @xmath45 such that any pair of distinct loops @xmath50 projects to distinct loops on @xmath4 and hence corresponds to distinct orbifold loops .
the @xmath49 obtained in this way patch together to a riemannian hilbert manifold @xmath51 by identifying elements that correspond to the same orbifold loop .
in fact , if distinct @xmath52 and @xmath53 are identified then , by the definition of the equivalence relation on developments , a whole open neighborhood of @xmath54 in @xmath55 is isometrically identified with an open neighborhood of @xmath56 in @xmath57 .
the free loop space @xmath58 is the metric completion of @xmath51 .
it is a riemannian hilbert orbifold with singular set @xmath59 corresponding to constant loops at the singular points of @xmath4 , cf .
section [ app : loop_space ] . given an atlas of @xmath4 the free loop space
@xmath29 can written as a quotient of a riemannian hilbert manifold @xmath60 , where @xmath61 is the disjoint union of the manifold charts of the atlas , and this description provides local manifold charts for @xmath29 . for a detailed description of the orbifold structure of @xmath29
we refer the reader to @xcite ( cf .
appendix [ app : loop_space ] ) . on @xmath29 the energy function @xmath62
is defined and its critical points correspond to the closed geodesics . since all singular points of @xmath29 have zero energy an explicit knowledge of their structure will not be relevant for our argument ( cf .
section [ appendix ] ) . for some @xmath63
we write @xmath64 and for a geodesic loop @xmath13 with @xmath65 we set @xmath66 . the spaces @xmath60 and @xmath29 admit finite dimensional approximations similarly as in the manifold case , see section [ app : loop_space ] . from our description of @xmath29
it is clear that the index @xmath67 and the nullity @xmath68 of a nontrivial orbifold geodesic can be defined as in the manifold case .
moreover , it shows that the statements used in @xcite on the index and the nullity of iterated geodesics and on local loop space homology remain valid in the following form since their proofs involve only local arguments .
note that there is a natural @xmath69-action on @xmath29 given by reparametrization .
[ lem : loop_local_structure_prop ] for a riemannian orbifold with isolated singularities and a nontrivial orbifold geodesic @xmath13 on it the following statements hold true . 1 .
either @xmath70 for all @xmath71 or @xmath72 grows linearly in @xmath71 ( * ? ? ?
* lemma 1 ) .
2 . there are positive integers @xmath73 and a sequence @xmath74 , @xmath75 , @xmath76 , such that the numbers @xmath77 are mutually distinct , @xmath78 , @xmath79 , and @xmath80 ( * ? ? ?
* lemma 2 ) .
3 . there exists some @xmath81 such that @xmath82 except possibly for @xmath83 ( * ? ? ?
* corollary 1 ) .
we will need the following slight generalizations of statements in @xcite .
the proofs are essentially the same as in @xcite . for convenience
we summarize the arguments in the appendix , section [ appendix ] . recall that a geodesic @xmath13 is called _ homologically invisible _ if @xmath84 .
[ thm : invisible_img ] let @xmath4 be a compact orbifold with isolated singularities and let @xmath13 be a closed geodesic on @xmath4 such that @xmath85 for all @xmath86 , i.e. @xmath13 does not have conjugate points when defined on @xmath87 .
suppose @xmath13 is neither homologically invisible nor an absolute minimum of @xmath62 in its free homotopy class .
then there exist infinitely many closed geodesics on @xmath4 .
note that if @xmath4 is simply connected , then a nontrivial geodesic @xmath13 is never an absolute minimum in its free homotopy class .
[ lem : img ] let @xmath4 be a compact orbifold with isolated singularities and let @xmath88 be a sequence of pairwise disjoint critical orbits such that the @xmath89 are not absolute minima of @xmath62 in their free homotopy classes .
suppose there exists @xmath90 such that @xmath91 for all @xmath92 .
then there exist infinitely many closed geodesics on @xmath4 .
note that the geodesics @xmath89 in the lemma need not be geometrically distinct .
the _ curve - shortening flow _ can be used to prove the existence of a simple closed geodesic on any riemannian @xmath0-sphere @xcite . in this section
we discuss properties of the curve - shortening flow on riemannian @xmath0-orbifolds that allow us to prove the existence of a _ separating geodesic _ on every simply connected riemannian spindle orbifold . here a loop embedded in the regular part of a spindle orbifold is called _ separating _ if each connected component of its complement contains at most one singular point .
let us first recall some well - known properties of the curve - shortening flow ( cf .
@xcite ) . for a smoothly embedded curve @xmath93 in a closed riemannian surface
there exists a unique maximal smooth curve - shortening flow @xmath94 for @xmath95 , @xmath96 , satisfying @xmath97 where @xmath81 is the curvature of @xmath98 and @xmath99 is its normal vector , which moreover depends continuously on the initial condition @xmath30 .
this flow can be considered as the negative gradient flow of the length functional .
an important feature of the curve - shortening flow is that it is a _ geometric flow _ meaning that the evolution of the geometric image of @xmath30 does not depend on the initial parametrization . in the situation above the curve @xmath98
is embedded for each @xmath95 . moreover ,
if @xmath100 is finite , then @xmath98 converges to a point .
if @xmath100 is infinite , then the curvature of @xmath98 converges to zero in the @xmath101-norm and a subsequence of @xmath98 converges to a closed embedded geodesic on @xmath8 . in particular
, @xmath100 is finite if the length of @xmath30 is sufficiently small ( * ? ? ?
if @xmath8 is not complete but the curvature is still bounded there is a single other alternative for finite @xmath100 namely if points on @xmath30 do not have a limit on @xmath8 for @xmath102 . in particular , we see that the curve - shortening flow of an embedded curve in the regular part of a compact @xmath0-orbifold is ( a priori ) defined until the flow hits a singular point or collapses to a point .
in fact , if the @xmath0-orbifold has only isolated singularities more is true as will be discussed below .
one possibility to analyse the local behaviour of a curve - shortening flow @xmath94 on a manifold at a space - time point @xmath103 , @xmath104 , is to blow up the flow at @xmath103 by a sequence of parabolic rescalings @xmath105 , @xmath106 , where @xmath107 denotes the riemannian metric .
such a blow - up sequence subconverges to a self - shrinking tangent flow of an embedded curve on @xmath108 ( cf .
@xcite ) which , according to a result of abresch and langer @xcite , is either a self - shrinking circle or a static straight line through the origin .
in the first case @xmath98 converges to @xmath7 as a `` round point '' .
another feature of the curve - shortening flow is that it satisfies the so - called _ avoidance principle _ , meaning that two initially disjoint curves remain disjoint under the flow .
moreover , by the strong maximum principle for parabolic pde it is impossible for a closed curve to be disjoint from a ( possible noncompact ) geodesic for @xmath109 and touch it tangentially at @xmath110 .
the following statement says that the curve - shortening flow can not spread out arbitrarily in finite time .
[ prp : area_rate ] if @xmath111 , then for every @xmath112 there exists @xmath113 , and an open set @xmath6 such that @xmath6 contains every @xmath114 , @xmath115 , and @xmath6 is contained in the @xmath116-neighborhood of each @xmath114 , @xmath115 .
now we analyse the evolution under the curve - shortening flow of a separating loop on a simply connected spindle orbifold .
[ prp : flow_orbifold ] let @xmath117 be a riemannian spindle orbifold , let @xmath30 be a separating loop on @xmath4 and let @xmath118 be the evolution of @xmath30 under the curve - shortening flow in the regular part of @xmath4 .
if @xmath119 hits a singular point @xmath7 of order @xmath120 in finite time @xmath100 , then the flow converges to this point . by lemma [ prp : area_rate ]
we can assume that there exists a point @xmath121 on @xmath4 which is avoided by @xmath119 such that each @xmath98 separates @xmath7 and @xmath122 .
if @xmath4 has two singular points , we choose @xmath122 to be a singular point .
the open subset @xmath123 of @xmath4 admits a @xmath24-fold manifold covering @xmath8 with a cyclic group of deck transformations @xmath9 of order @xmath24 that acts by rotations around the preimage @xmath124 of @xmath7 in @xmath8 .
the preimages @xmath125 of @xmath114 in @xmath8 are embedded @xmath9-invariant circles and are solutions of the curve - shortening flow . by parabolically blowing up this flow at @xmath126
as discussed above we obtain a @xmath9-invariant tangent flow . because of @xmath127 this tangent flow must be a circle and so @xmath125 converges to @xmath124 ( cf . the discussion preceding lemma [ prp : area_rate ] ) .
in particular , @xmath98 converges to @xmath7 .
[ prp : flow_orbifold_gen ] let @xmath117 be a simply connected riemannian spindle orbifold , let @xmath30 be a separeting loop on @xmath4 and let @xmath118 be the maximal evolution of @xmath30 under the curve - shortening flow in the regular part of @xmath4 .
then the curve @xmath30 either 1
. shrinks to a round point in the regular part of @xmath4 in finite time or 2 .
collapses into a singular point of @xmath4 in finite time or 3 .
@xmath128 and @xmath30 stays in the regular part of @xmath4 forever .
not that the first case can only occur if @xmath4 has at most one singular point .
we only have to exclude the case that @xmath119 hits a singular point @xmath7 of @xmath4 in finite time @xmath129 without collapsing into this point .
suppose this were the case .
by lemma [ prp : flow_orbifold ] we can assume that the order @xmath24 of @xmath7 is even and that the flow does not approach a singular point of odd order at time @xmath100 .
since @xmath24 and @xmath25 are coprime by assumption , the order @xmath25 must be odd .
we can assume that there exists a point @xmath121 on @xmath4 such that the flow avoids a whole neighborhood of @xmath122 , such that each @xmath98 separates @xmath7 and @xmath122 and such that @xmath130 lies in the regular part of @xmath4 .
in fact , for @xmath131 we can choose @xmath122 to be the singular point of odd order @xmath25 and otherwise we can apply lemma [ prp : area_rate ] .
as in the proof of lemma [ prp : flow_orbifold ] the open subset @xmath123 of @xmath4 admits a @xmath24-fold manifold covering @xmath8 with a cyclic group of deck transformations @xmath9 of order @xmath24 that acts by rotations around the preimage @xmath124 of @xmath7 in @xmath8 . also , the preimages @xmath125 of @xmath114 in @xmath8 are embedded @xmath9-invariant circles and are solutions of the curve - shortening flow . since @xmath119 avoids a neighborhood of @xmath122 and since @xmath125 does not converge to a point at time @xmath100 , the flow @xmath125 can be extended to a flow @xmath132 with @xmath133 .
now the fact that @xmath119 hits @xmath7 at time @xmath100 implies that the extended flow @xmath134 develops a self - crossing at time @xmath100 .
this contradicts embeddedness and hence the claim follows . using compactness and
the fact that the curvature converges to zero in infinite time , one proves as in the manifold case [ lem : sub_to_min_geo ] in the situation of the preceding proposition , case @xmath135 , the loop subconverges to a nontrivial ( orbifold ) geodesic .
note that the limit geodesic obtained in lemma [ lem : sub_to_min_geo ] is a priori not necessarily contained in the regular part of @xmath4 .
[ prp : simple_geodesic ] on every simply connected riemannian spindle orbifold @xmath117 there exists a separating geodesic .
in particular , there exists a closed geodesic in the regular part of any spindle orbifold .
the first statement is optimal in the sense that there exist riemannian metrics on @xmath23 all of whose geodesics are closed but with only one embedded geodesic @xcite .
choose distinct @xmath136 such that @xmath130 is contained in the regular part of @xmath4 .
we smoothly foliate @xmath130 by circles separating @xmath7 and @xmath122 . then , under the curve - shortening flow small circles near @xmath7 flow to a point in finite time , and so do small circles near @xmath122 ( * ? ? ?
7.1 ) . however , the orientations of the limiting points will be opposite in both cases .
hence , by continuous dependence of the flow on the initial conditions there must be some circle @xmath30 in the middle that does not flow to a point in finite time , but instead stays in the regular part of @xmath4 forever by proposition [ prp : flow_orbifold ] . by lemma [ lem : sub_to_min_geo ] this circle subconverges to some orbifold geodesic @xmath13 .
it remains to show that @xmath13 is contained in the regular part .
in fact , in this case @xmath13 is embedded and separating as a limit of embedded and separating loops .
let @xmath137 be a subsequence of the curve - shortening flow @xmath98 that converges to @xmath13 .
suppose that @xmath13 hits a singular point , say @xmath7 of order @xmath24 .
we can assume that the @xmath137 and @xmath13 avoid a neighborhood of a point @xmath138 on @xmath4 .
the open subset @xmath139 of @xmath4 admits a @xmath24-fold orbi - cover @xmath140 with a cyclic group of deck - transformations @xmath9 of order @xmath24 that acts by rotations around the preimage @xmath124 of @xmath7 in @xmath140 .
let @xmath141 be a sequence such that @xmath142 converges to some @xmath143 , @xmath144 .
the restrictions of @xmath137 to @xmath145 and of @xmath13 to @xmath146 can be lifted to embedded curves @xmath147 , @xmath148 , and to geodesics @xmath149 , @xmath148 , that are permuted by the deck transformation group @xmath9 .
we can choose the @xmath150 , @xmath151 and the @xmath152-numbering in such a way that @xmath153 converges to @xmath154 .
note that for fixed @xmath155 the @xmath153 have disjoint images in @xmath140 since @xmath137 is embedded in @xmath4 . for @xmath120
this yields a contradiction since the geodesics @xmath154 intersect transversally at @xmath124 in this case .
suppose that @xmath156 .
in this case @xmath8 is rotated about @xmath2 around @xmath124 by the deck transformation group and the orbifold geodesic @xmath13 is reflected at the singular point @xmath7 on @xmath4 .
the only way for @xmath13 to reverse its direction is by being reflected at a singular point of even order .
hence , by periodicity it has to be reflected at singular points of even order twice during a single period .
since @xmath24 and @xmath25 are coprime by assumption , this second reflection also has to occur at @xmath7 .
moreover , this second reflection has to occur from a different direction , because otherwise there would have to be an additional encounter with the singularity of even order in between .
therefore , the @xmath157 have transverse self - intersections at @xmath124 .
now , in this case the transversality argument from the case @xmath158 yields a contradiction since the @xmath159 are embedded and hence the claim follows .
as seen in section [ sub : riem_orb ] it is sufficient to prove our main result for simply connected spindle orbifolds . in view of our results from the preceding section , in this case
a proof can be given similarly as in @xcite in the case of a @xmath0-sphere as we will discuss now .
let @xmath117 be a simply connected riemannian spindle orbifold .
by lemma [ prp : simple_geodesic ] there exists a separating geodesic @xmath160 .
suppose the following two conditions are satisfied . for every geodesic @xmath161 with initial point @xmath162 on @xmath163
there exists @xmath164 with @xmath165 .
when we consider @xmath13 as defined on @xmath87 there exists a pair of conjugate points of @xmath13 .
note that the second statement is equivalent to the condition that for every @xmath166 there exists @xmath167 such that @xmath168 and @xmath169 are conjugate points of @xmath30 ( cf .
* cor . 1.3 ) ) . in this case
birkhoff s annulus map @xmath170 can be defined on the closed annulus @xmath171 $ ] as follows ( cf .
@xcite and ( * ? ? ?
* vi . 10 ) ) . for @xmath172 and @xmath173
consider a geodesic @xmath30 starting at @xmath174 in a direction that forms an angle of @xmath175 with @xmath176 .
condition @xmath177 guarantees the existence of some time @xmath178 at which @xmath30 returns to an encounter with @xmath13 for the second time , say at @xmath179 .
then @xmath180 and @xmath179 enclose some angle @xmath181 and one sets @xmath182 . in case of @xmath183 a second conjugate point can be used to extend the map @xmath170 to all of @xmath171 $ ] in a continuous way .
the map @xmath170 is isotopic to the identity and preserves a canonical area measure related to the liouville measure on the unit tangent bundle of @xmath184 which is invariant under the geodesic flow .
moreover , the restrictions of @xmath170 to the two boundary components are inverse to each other . in this case
the work of franks @xcite implies that birkhoff s annulus map has infinitely many periodic points ( cf .
* thm . 4.1 ) ) and these points correspond to closed geodesics on @xmath4 . in the following
we show that the existence of infinitely many geodesics can still be shown if birkhoff s annulus map can not be defined .
[ prp : separating_without_conjugate_points ] suppose on a riemannian spindle orbifold @xmath117 there exists a separating geodesic @xmath13 without conjugate points . then there
exist infinitely many closed geodesics .
the proof works similarly as the proof of ( * ? ? ?
* thm . 1 ) .
we choose riemannian manifold charts @xmath185 , @xmath186 , @xmath187 of @xmath4 where @xmath188 and where @xmath186 and @xmath187 cover @xmath123 and @xmath189 , respectively .
moreover , we choose finite - dimensional approximations @xmath190 and @xmath191 of @xmath192 ( cf .
section [ app : loop_space ] ) and @xmath193 containing @xmath194 as described at the end of section [ app : loop_space ] .
since @xmath13 is separating , there exists a tubular neighborhood @xmath195 of @xmath163 in the regular part of @xmath4 which is homeomorphic to an annulus . for every @xmath196 we let @xmath197 denote the set of @xmath198 which have energy @xmath199 and whose projections to @xmath4 lie in @xmath195 and are freely homotopic to @xmath13 in @xmath195 .
since we are looking for infinitely many closed geodesics we may assume that the only closed geodesics freely homotopic to @xmath13 in @xmath195 are those in @xmath194 .
moreover , we can choose @xmath195 so small that every @xmath200 with @xmath201 is disjoint from @xmath163 .
this follows from the assumption that @xmath13 does not have conjugate points and the gau lemma , cf .
* thm . 1 ) . if we choose arbitrarily small @xmath195 and @xmath116 the sets @xmath202 form a fundamental system of neighborhoods of @xmath194 in @xmath203 .
therefore , we have that either ( cf .
* thm . 1 ) ) 1 .
@xmath13 is a local minimum of @xmath62 or 2 .
@xmath13 can be approximated by curves @xmath198 with @xmath201 from both sides or 3 .
@xmath13 can be approximated by curves @xmath198 with @xmath204 precisely from one side . in the second case it follows as in (
1 ) that @xmath205 and this implies the existence of infinitely many closed geodesics by theorem [ thm : invisible_img ] . in case @xmath177 or @xmath135
let @xmath206 be a disk bounded by @xmath163 such that @xmath13 can not be approximated by closed curves in @xmath206 with @xmath204 and suppose that @xmath207 has order @xmath24 .
the disk @xmath206 is @xmath24-foldly covered by a disk @xmath208 in @xmath186 .
a parametrization @xmath209 of the boundary of @xmath208 by arc - length is a geodesic that covers @xmath13 @xmath24-times .
we can choose finite - dimensional approximations @xmath210 of @xmath60 containing a homotopy in @xmath208 from @xmath211 to a point curve .
as above we choose an annulus @xmath212 so small that @xmath194 are the only closed geodesics freely homotopic to @xmath13 in @xmath195 . moreover , we may assume that every @xmath200 which lies in the component @xmath213 of @xmath214 contained in @xmath206 has energy @xmath215 .
hence we have @xmath216 for all @xmath217 which are contained in @xmath206 , because otherwise such a @xmath30 with @xmath218 would be a closed geodesic freely homotopic to @xmath13 .
the analogous statement also holds for @xmath219 @xcite .
in particular , the analogous statement also holds for @xmath220 . in this very situation
min - max methods applied to homotopies in @xmath208 from @xmath211 to a point curve are used in the proof of ( * ? ? ?
* thm . 1 ) to show the existence of a sequence of closed geodesics @xmath221 in @xmath208 such that @xmath222 tends to infinity and such that the local groups @xmath223 do not vanish .
the gradient of the energy functional restricted to the finite - dimensional approximation is used to deform the homotopies and so the fact that @xmath208 is bounded by a geodesic guarantees that the construction remains in @xmath208
. the resulting geodesics project to ( orbifold ) geodesics @xmath224 in @xmath206 with @xmath225 tending to infinity and with @xmath226 . therefore there exist infinitely many closed geodesics on @xmath4 by theorem [ thm : invisible_img ] . in this section
we study the case of a riemannian spindle orbifold with a separating geodesic for which the corresponding birkhoff map is not defined .
we reduce the existence of infinitely many closed geodesics to theorem 1 .
finally we explain how this implies our main result .
the following lemma is a special case of ( * ? ? ?
* lem . 2 ) .
[ lem : without_conj ] suppose @xmath13 is a separating geodesic with conjugate points on a riemannian spindle orbifold
. then @xmath13 can be approximated from either side by closed curves @xmath30 which are disjoint from @xmath13 and satisfy @xmath201 . in particular , @xmath13 can be approximated from either side by shorter , disjoint curves .
now we can prove [ prp : separating_with_conjugate_points ] suppose @xmath13 is a separating geodesic with conjugate points on a simply connected riemannian spindle orbifold @xmath117 and @xmath227 is a geodesic disjoint from @xmath13 .
then there exist infinitely many closed geodesics .
the proof is similar as the proof of ( * ? ? ?
* thm . 2 ) .
we use the curve - shortening flow instead of birkhoff s curve shortening process and simplify the second part of the argument .
let @xmath206 be the component of @xmath228 that contains @xmath229 .
since @xmath13 is separating by assumption , there exists an open neighborhood @xmath195 of the closure of @xmath206 in @xmath4 that admits a riemannian manifold chart @xmath230 .
the geodesic @xmath231 lifts to a geodesic @xmath232 on @xmath230 and the geodesic @xmath13 is covered @xmath24-times by a geodesic @xmath209 disjoint from @xmath232 . in (
2 ) it is proven that the closure of every limit geodesic @xmath233 of @xmath232 , that is every geodesic of the form @xmath234 , @xmath235 where @xmath236 is an accumulation point of @xmath237 in @xmath238 , is disjoint from @xmath239 . in particular , the closure of the image @xmath240 in @xmath195 of such a limit geodesic of @xmath232 is disjoint from @xmath163 .
let @xmath241 be the component of @xmath242 that contains @xmath163 in its closure .
since the closure of @xmath243 is disjoint from @xmath163 , by lemma [ lem : without_conj ] there is an embedded loop @xmath244 in @xmath241 that is freely homotopic in the regular part of @xmath241 to @xmath13 and shorter than @xmath13 .
we claim that the evolution @xmath98 of @xmath244 under the curve - shortening flow does not leave @xmath241 .
otherwise there existed some @xmath245 minimal with the property that @xmath246 is not contained in @xmath241 .
let @xmath247 . by the avoidance principle applied to @xmath13 and @xmath244 the only possibility could be that @xmath7 is contained in the closure of @xmath243 .
since @xmath244 is non - contractible in @xmath241 and since @xmath243 is not a point , the point @xmath7 is regular by lemma [ prp : flow_orbifold ] and so is @xmath248 by lemma [ prp : flow_orbifold_gen ] .
let @xmath249 be the geodesic which is tangent to @xmath248 at @xmath7 .
by minimality of @xmath245 the geodesic @xmath250 is contained in the closure of @xmath243 . in particular ,
the flow of @xmath244 and the ( static ) flow of @xmath250 touch at @xmath251 for the first time .
this is impossible by the maximum principle ( cf .
section [ sec : curve_shortening ] ) and hence the evolution of @xmath244 stays in @xmath241 as claimed . moreover , since @xmath244 is non - contractible in @xmath241 by assumption , it evolves in the regular part forever , i.e. we are in case @xmath135 of lemma [ prp : flow_orbifold ] . by lemma [ lem : sub_to_min_geo ] the flow subconverges to a simple closed geodesic @xmath252 contained in the closure of @xmath241 .
the proof of proposition [ prp : simple_geodesic ] shows that this limit geodesic actually lies in the regular part of @xmath4 .
it is distinct from @xmath13 since @xmath244 is shorter than @xmath13 and the curve - shortening flow does not increase the arclength . by the choice of @xmath244 the geodesic @xmath252 is separating and so we are done in this case by proposition [ prp : separating_without_conjugate_points ] if @xmath252 does not have conjugate points .
otherwise we define @xmath253 to be the component of @xmath254 whose closure contains @xmath13 .
this component is bounded by two geodesics and contained in the regular part of @xmath4 by the construction of @xmath252 .
moreover , it contains a non - contractible embedded loop @xmath255 which is shorter than @xmath252 by lemma [ lem : without_conj ] and hence also shorter than @xmath13 . again , letting @xmath255 evolve under the curve - shortening flow the same argument as above yields a separating limit geodesic @xmath256 in @xmath257 which is now distinct from both @xmath13 and @xmath252 and hence contained in @xmath253
. this process can be iterated .
it either yields infinitely many ( simple ) closed geodesics on @xmath4 with conjugate points or terminates at a separating geodesic without conjugate points which in turn implies the existence of infinitely many closed geodesics by proposition [ prp : separating_without_conjugate_points ] . from what has been said in section [ sub : outline ] we see that propositions [ prp : separating_with_conjugate_points ] and [ prp : separating_without_conjugate_points ] imply the following statement .
let @xmath13 be a separating geodesic on a riemannian spindle orbifold @xmath117 for which birkhoff s annulus map @xmath170 is not defined . then there exist
infinitely many closed geodesics .
recall from section [ sub : outline ] that frank s work implies the existence of infinitely many closed geodesics in the case in which birkhoff s annulus map @xmath170 can be defined ( cf .
section [ sub : outline ] ) .
therefore , in any case there are infinitely many closed geodesics on a riemannian spindle orbifold . by the remark at the end of section [ sub : riem_orb ] our main result follows .
we summarize the description of an orbifold loop space from @xcite . for
a riemannian orbifold @xmath4 let @xmath61 be the disjoint union of riemannian manifold charts covering @xmath4 and let @xmath258 be the small category with set of objects @xmath61 and arrows being the germs of change of charts of @xmath61 .
then @xmath259 with the usual topology of germs on @xmath258 is an tale groupoid and @xmath4 can be represented as @xmath260 ( * ? ? ?
@xmath258-loop based at @xmath261 _ over a subdivision @xmath262 of the interval @xmath263 $ ] is a sequence @xmath264 where 1 .
@xmath265 { \rightarrow}x$ ] is of class @xmath32 , i.e. @xmath89 is absolutely continuous and the velocity functions @xmath266 are square integrable .
@xmath267 is an element of @xmath258 such that @xmath268 for @xmath269 , @xmath270 for @xmath271 and @xmath272 . here
@xmath273 and @xmath274 denote the source and the target of @xmath275 ( cf .
. a @xmath258-loop is called _ geodesic _ if all the @xmath89 are geodesics and their velocities match up at the break points @xmath276 via the @xmath267 .
a geodesic @xmath258-loop gives rise to a closed geodesic on @xmath4 in the sense of definition [ dfn : orb_geo ] .
the space @xmath277 is defined as the set of equivalence classes of @xmath258-loops based at @xmath7 under the equivalence relation generated by the following operations ( * ? ? ?
* sect . 2.3.2 , 3.3.2 )
: 1 . given a @xmath258-loop @xmath264 over the subdivision @xmath278
, we can add a subdivision point @xmath279 together with the unit element @xmath280 to get a new sequence , replacing @xmath89 in @xmath13 by @xmath281 , @xmath282,@xmath283 , where @xmath281 and @xmath283 are the restrictions of @xmath89 to the intervals @xmath284 $ ] and @xmath285 $ ] and where @xmath286 , @xmath287 is the germ of the identity map at @xmath122 .
2 . replace a @xmath258-loop @xmath13 by a new loop @xmath288 over the same subdivision as follows : for each @xmath271 choose @xmath32-maps @xmath289 { \rightarrow}\mathcal{g}$ ] such that @xmath290 , and define @xmath291 , @xmath292 for @xmath293 , @xmath294 and @xmath295 . the space @xmath60 is defined to be @xmath296 .
it admits a natural structure of a riemannian hilbert manifold ( * ? ? ?
a @xmath258-loop @xmath13 defines a development @xmath297 in the sense of definition [ dfn : develop ] .
a neighborhood of the equivalence class of @xmath13 in @xmath60 is isometric to a neighborhood of @xmath33 in @xmath45 . if @xmath4 is compact , then @xmath60 is complete with respect to the induced metric ( cf .
proof of ( * ? ? ?
1.4 ) ) . an energy function can be defined on @xmath60 whose critical points correspond to geodesic @xmath258-loops ( * ? ? ?
the groupoid @xmath258 acts isometrically on the left on @xmath60 .
if @xmath298\in \omega_x$ ] is represented by the @xmath258-loop @xmath264 based at @xmath7 and @xmath107 is an element of @xmath258 with @xmath299 and @xmath300 , then @xmath301 $ ] is represented by @xmath302 ( * ? ? ?
the quotient @xmath303 is called the _ free loop space _ of @xmath4 .
the quotient map @xmath304 induces a natural structure of a riemannian hilbert orbifold on @xmath305 noted @xmath29 ( * ? ? ? * sect .
since @xmath60 is complete as a metric space , so is @xmath29 .
the space @xmath306 is naturally a subset of @xmath29 and coincides with the ordinary loop space of @xmath307 as a riemannian manifold ( cf .
if @xmath4 has only isolated singularities , then the only elements of @xmath60 with nontrivial @xmath258-isotropy are the loops that project to a singular point of @xmath4 . in this case
@xmath258-equivalence classes of @xmath258-loops correspond to equivalence classes of developments discussed in section [ sub : orb_loop ] and @xmath29 can be described as in section [ sub : orb_loop ] . 1 .
there exist @xmath311 and a big enough @xmath81 such that every element of @xmath192 can be represented by a @xmath258-path @xmath264 defined over the subdivision @xmath312 , where @xmath313 , and each @xmath314 is the center of a convex open geodesic ball of radius @xmath315 containing the image of @xmath89 .
the space @xmath192 retracts by energy - nonincreasing deformation onto the subspace @xmath316 whose elements are represented by @xmath258-loops as above for which each @xmath89 is a geodesic segment .
3 . the restrictions of the energy function to @xmath192 and to @xmath316 have the same critical points , and at each such point the nullity and the index are the same .
4 . the orbifold @xmath317 retracts by energy - nonincreasing deformation onto the finite - dimensional orbifold @xmath318 . for a riemannian spindle orbifold @xmath319 with a seperating geodesic @xmath13 we can choose the following convenient charts . let @xmath136 be two points in different components of the complement of @xmath13 such that @xmath130 lies in the regular part of @xmath4 .
we choose @xmath61 to be the disjoint union of @xmath320 and riemannian manifold charts @xmath186 and @xmath187 covering @xmath123 @xmath24-foldly and @xmath189 @xmath25-foldly , respectively .
with respect to this choice any @xmath258-loop based at @xmath321 that projects to @xmath307 can be represented as @xmath322 for an @xmath32-loop in @xmath323 with @xmath324 and can thus be identified with @xmath13 .
any @xmath258-equivalence class of @xmath258-loops based at @xmath325 which is represented by a @xmath258-loop that projects to @xmath123 can be represented as @xmath326 for an @xmath32-loop in @xmath323 with @xmath327 and @xmath328 and a deck transformation @xmath329 of the covering @xmath330 . in this section
we explain why the proofs of ( * ? ? ?
* thm . 3 ) and ( * ? ? ?
2 ) also work in the slightly more general situation of theorem [ thm : invisible_img ] and lemma [ thm : loop_iteration ] which allow isolated orbifold singularities .
the first step is to obtain ( * ? ? ?
1 ) in the following version . here
the map @xmath331 sends a loop @xmath30 to its @xmath71-fold iteration @xmath332 .
[ thm : loop_iteration ] let @xmath333 be the free loop space of a compact riemannian orbifold @xmath4 with isolated singularities . for some @xmath334 let @xmath335 $ ] be a class in @xmath336
. then @xmath337:=[\psi^m \circ g ] \in \pi_p(|\lambda|,|\lambda^{\kappa m^2}|)$ ] is trivial for almost all @xmath86 . in the manifold case
the idea of the proof is the following . given a homotopy @xmath338{\rightarrow}\lambda$ ] , then @xmath339 is a naturally associated homotopy from @xmath340 to @xmath341 .
one can replace this homotopy , which pulls the @xmath71 loops of @xmath340 to @xmath341 as a whole , by another homotopy @xmath342{\rightarrow}\lambda$ ] from @xmath343 to @xmath344 that pulls the @xmath71 loops from @xmath340 to @xmath341 successively .
the advantage of the new homotopy over the old one is that the energy of @xmath345 depends only on @xmath346 and @xmath347 in the limit of large @xmath71 and can thus be bounded appropriately .
this construction can be applied fiberwise to a map @xmath348 with respect to a splitting @xmath349 and yields a homotopic map with image in @xmath350 .
the same construction can be carried out in the case of an orbifold with isolated singularities .
in fact , all regularity issues occurring in @xcite can be handled in the same way in this case since finite - dimensional approximations are also available for orbifold loop spaces ( see proposition [ prp : finite_approx ] ) and since the whole construction can be assumed to take place away from the energy zero level set and hence in the manifold part of @xmath333 .
[ thm : homot_to_homol ] let @xmath333 be the free loop space of a compact riemannian orbifold @xmath4 with isolated singularities .
let @xmath351 be an element of @xmath352 , where @xmath334 and @xmath353 is the union of all components of @xmath333 intersecting @xmath354 .
let @xmath355 be a finite set of integers @xmath356 .
then there exists @xmath15 such that no @xmath357 divides @xmath71 and such that @xmath358 vanishes in @xmath359 .
the proof of ( * ? ? ?
* theorem 2 ) can be taken verbatim as a proof for theorem [ thm : homot_to_homol ] .
for sufficiently large @xmath71 one can construct a homotopy from a representative of @xmath358 to a representative in @xmath360 by using theorem [ thm : loop_iteration ] inductively .
now we explain the proof of ( * ? ? ?
* theorem 3 ) and why it generalizes to the setting of theorem [ thm : invisible_img ] .
suppose that there exist only finitely many towers of closed geodesics on @xmath4 .
then all critical @xmath69-orbits on @xmath333 are isolated .
moreover , by lemma [ lem : loop_local_structure_prop ] , @xmath135 , and perhaps after choosing a different @xmath13 , one can find @xmath361 such that @xmath362 and @xmath363 for every @xmath364 and every closed geodesic @xmath231 with @xmath365 for all @xmath71 . in the manifold case lemma [ lem : loop_local_structure_prop ] ,
@xmath366 , implies that there exist integers @xmath367 , @xmath368 , such that @xmath369 is an isomorphism whenever none of the @xmath370 divides @xmath71 ( for details see the proof of ( * ? ? ?
* theorem 2 ) ) .
the same conclusion holds in the present situation since its proof involves only local arguments . by lemma [ lem : loop_local_structure_prop ] , @xmath177 , and the assumption of only finitely many towers of closed geodesics there exists some @xmath371 such that every closed geodesic @xmath231 with @xmath372 either satisfies @xmath373 or @xmath365 for all @xmath71
hence one has @xmath374 whenever @xmath231 is a closed geodesic with @xmath372 .
therefore , as in @xcite , standard arguments from morse theory imply that @xmath375 is one - to - one if @xmath376 .
in particular , the composition @xmath377 is one - to - one .
since @xmath13 is not an absolute minimum in its free homotopy class , this contradicts theorem [ thm : homot_to_homol ] . hence theorem [ thm : invisible_img ] holds .
the proof of ( * ? ? ?
* lemma 2 ) and lemma [ thm : loop_iteration ] works as follows .
@xmath378 implies @xmath379 by lemma [ lem : loop_local_structure_prop ] , @xmath135 .
since for every closed geodesic @xmath13 either @xmath72 grows linearly with @xmath71 or @xmath70 for all @xmath71 , the @xmath89 can be iterates of a finite number of prime closed geodesics only if @xmath380 for some @xmath155 and all @xmath71 . in this case theorem [ thm : invisible_img ] proves the existence of infinitely many closed geodesics .
_ acknowledgements .
_ i would like to thank nick edelen and alexander lytchak for discussions and comments on a draft version of this paper .
i also thank victor bangert and stephan wiesendorf for remarks and hints to the literature , respectively .
g. huisken and a. polden , geometric evolution equations for hypersurfaces , in _ calculus of variations and geometric evolution problems ( cetraro , 1996 ) _ , 4584 , lecture notes in math .
, 1713 , fond .
cime / cime found .
subser , springer , berlin . | we show that on every compact riemannian @xmath0-orbifold there exist infinitely many closed geodesics of positive length . |
the ability to simulate the final inspiral , merger , and ring - down of black hole binaries with numerical relativity @xcite plays a key role in understanding a source of gravitational waves that may one day be observed with gravitational wave detectors . while initial simulations focused on binaries of equal - mass , zero spin , and quasi - circular inspirals
, there currently is a large effort to explore the parameter space of binaries , e.g. @xcite .
a key part of studying the parameter space is to simulate binaries with intermediate mass - ratios . to date , the mass ratio furthest from equal masses that has been numerically simulated is 10:1 @xcite .
these simulations use the baumgarte - shapiro - shibata - nakamura ( bssn ) formulation @xcite with 1+log slicing , and the @xmath0 driver condition for the shift @xcite . in @xcite
, it was noted that the stability of the simulation is sensitive to the damping factor , @xmath1 , used in the @xmath0 driver condition , @xmath2 here , @xmath3 is the shift vector describing how the coordinates move inside the spatial slices , @xmath4 , and @xmath5 is the contraction of the christoffel symbol , @xmath6 , with the conformal metric , @xmath7 . the standard choice for @xmath1 is to set it to a constant value , which works well even for the most demanding simulations as long as the mass ratio is sufficiently close to unity . in binary simulations , a typical choice is a constant value of about @xmath8 , with @xmath9 the total mass of the system .
this choice , however , leads to instabilities for the mass ratio 10:1 simulation @xcite , although stability was obtained for @xmath10 .
the value of @xmath1 is chosen to damp an outgoing change in the shift while still yielding stable evolutions . as we will show , if @xmath1 is too small , there are unwanted oscillations , and values that are too large lead to instabilities . by itself , this observation is not new , see e.g. @xcite .
the key issue for unequal masses is that , as evident from ( [ eq : gammadriver ] ) , the damping factor @xmath1 has units of inverse mass , @xmath11 .
therefore , the interval of suitable values for @xmath1 depends on the mass of the black holes . for unequal masses , a constant @xmath1 can
not equally well accommodate both black holes .
a constant damping parameter implies that the effective damping near each black hole is asymmetric since the damping parameter has dimensions @xmath11 . for large mass ratios
, this asymmetry in the grid can be large enough to lead to a failure of the simulations because the damping may become too large or too small for one of the black holes . to cure this problem , we need a position - dependent damping parameter that adapts to the local mass
. in particular , we want it to vary such that , in the vicinity of the @xmath12 puncture with mass @xmath13 , its value approaches @xmath14 .
a position - dependent @xmath1 was already considered when the @xmath0 driver condition was introduced @xcite , but such constructions were not pursued further because for moderate mass ratios a constant @xmath1 works well .
recently , we revived the idea of a non - constant @xmath1 for moving puncture evolutions in order to remove the limitations of a constant @xmath1 for large mass ratios . in @xcite , we constructed a position - dependent @xmath1 using the the conformal factor , @xmath15 , which carries information both about the location of the black holes , and about the local puncture mass .
the form of @xmath1 was chosen to have proper fall - off rates both at the punctures and at large distance from the binary . in @xcite ,
this approach was used successfully for mass ratio 10:1 .
( we note in passing that damping is useful in other gauges as well , e.g. in @xcite the modified harmonic gauge condition includes position - dependent damping by use of the lapse function . ) in the present work , we examine one potential short - coming of the choice of @xcite , which leads us to suggest an alternative type of position - dependent @xmath1 . using @xcite , we find large fluctuations in the values of that @xmath1 , and this might lead to instabilities in the simulation of larger mass - ratio binary black holes . to address this ,
we have tested two new explicit formulas for the damping factor designed to have predictable behavior throughout the domain of computation .
we find the new formulas to produce only small changes in the waveforms that diminish with resolution , and there is a great deal of freedom in the implementation .
independently of our discussion here , in @xcite the stability issues for large @xmath1 are explained , and a non - constant @xmath1 is suggested ( although not yet explored in actual simulations ) , that , in its explicit coordinate dependence , is similar to one of our suggestions .
the paper is organized as follows .
we first describe the reasons for the damping factor and some of the reasons for limiting its value in sec [ sec : motivation ] . in sec .
[ sec : forms_of_eta ] , we discuss some previous forms of @xmath1 that have been used .
we also present two new definitions and why we investigated them . in sec .
[ sec : results ] , we find that these new definitions agree well with the use of constant @xmath1 in the extracted gravitational waves for mass ratios up to 4:1 . finally , in sec . [
sec : discussion ] , we discuss further implications of this work .
in order to define a position - dependent form for @xmath1 , it is important to determine what this damping parameter accomplishes in numerical simulations . for this reason , we examine the effects of running different simulations while varying @xmath1 between runs .
first we use evolutions of single non - spinning black - holes to identify the key physical changes .
then we examine equal - mass binaries to determine specific values desired in @xmath1 at both large and small radial coordinates .
for all the work in this paper , we have used the bam computer code described in @xcite .
it uses the bssn formalism with 1+log slicing and @xmath0 driver condition in the moving puncture framework @xcite .
puncture initial data @xcite with bowen - york extrinsic curvature @xcite have been used throughout this work , solving the hamiltonian constraint with the spectral solver described in @xcite . for binaries
, parameters were chosen using @xcite to obtain quasi - circular orbits , while the parameters for single black holes were chosen directly .
we extract waves via the newman - penrose scalar @xmath16 .
the wave extraction procedure is described in detail in @xcite .
we perform a mode decomposition using spin - weighted spherical harmonics with spin weight @xmath17 , @xmath18 , as basis functions and calculate the scalar product @xmath19 we further split @xmath20 into mode amplitude @xmath21 and phase @xmath22 in order to cleanly separate effects in these components , @xmath23 . in this paper , we focus on one of the most dominant modes , the @xmath24 mode , and report results for this mode unless stated otherwise . the extraction radius used here is @xmath25 .
the damping factor , @xmath1 , in eq .
( [ eq : gammadriver ] ) , is included to reduce dynamics in the gauge during the evolution . to examine the problem brought up in the introduction , we compare results of a single , non - spinning puncture with mass @xmath9 .
we use a courant factor of @xmath26 and 9 refinement levels centered around the puncture . the resolution on the finest grid
is @xmath27 , and the outer boundary is situated at @xmath28 .
varying the damping constant between @xmath29 and @xmath30 , two main observations can be made .
first , as designed , a non - zero @xmath1 attenuates emerging gauge waves efficiently .
second , an instability develops for values of @xmath1 that are too large . figs .
[ fig : spbetax1_t15.2 ] , and [ fig : spbetax1_t30.4 ] illustrate the first observation .
both figures show the @xmath31-component of the shift along the @xmath31-coordinate using @xmath32 .
-component of the shift , @xmath33 , for a single non - spinning puncture of mass @xmath9 at time @xmath34 .
the three lines were taken for different values of the damping factor @xmath1 .
the solid line ( black ) is for @xmath35 .
the dashed line ( red ) is for @xmath36 and the dotted - dashed line ( green ) is for @xmath37 .
this shows the beginning of a pulse in @xmath33 for smaller values of @xmath1 . ]
-component of the shift , @xmath33 , for a single non - spinning puncture of mass @xmath9 at time @xmath38 .
the three lines were taken for different values of the damping factor @xmath1 with the same line type and color scheme as in fig .
[ fig : spbetax1_t15.2 ] .
here it is clear a pulse radiates outward in the shift with smaller values of @xmath1 . ] apart from the usual shift profile , fig .
[ fig : spbetax1_t15.2 ] shows the beginnings of a pulse in the @xmath35 case ( solid line ) at @xmath39 after @xmath40 of evolution . examining fig .
[ fig : spbetax1_t30.4 ] , where we zoom in at a later time , @xmath41 , one can see that the pulse has started to travel further out ( solid line ) . looking carefully , one can also see a much smaller pulse in the @xmath36 line ( dashed ) .
lastly , by examination , one can find almost no traveling pulse in the @xmath37 curve ( dotted - dashed line ) .
the observed pulse in the shift travels to regions far away from the black hole and effects the gauge of distant observers .
this might have undesirable implications for the value of such numerical data when trying to understand astrophysical sources .
-component of the shift vector in the @xmath31-direction for a single non - spinning puncture of mass @xmath9 at times @xmath42 and @xmath41 .
the three different lines mark three values of the damping constant @xmath1 .
the solid line ( black ) is for @xmath43 , the dashed line ( red ) for @xmath44 and the dotted - dashed line ( green ) for @xmath45 . at @xmath46 ,
the simulation using @xmath45 develops an instability in the shift vector and fails soon afterward , the same happens for @xmath44 at @xmath41 . in the simulation using @xmath47 , no such instability develops ( not shown ) . ] for values of @xmath1 larger than @xmath48 , an instability arises in the shift at larger radius .
[ fig : betax_fail ] shows the @xmath31-component of the shift vector using damping constants @xmath43 ( solid line ) , @xmath44 ( dashed line ) and @xmath45 ( dotted - dashed line ) .
the plots show an instability in simulations with @xmath49 developing in @xmath3 , which eventually leads to a failure of the simulations .
contrary to this , the simulation using @xmath43 does not show this shift related instability . in test runs we found that by decreasing the courant factor used , we could increase the value of the damping factor and still get stable evolutions .
this agrees with @xcite where it was shown that the gamma driver possesses the stiff property , which limits the size of the time - step in numerical integration based on the value of the damping .
figures [ fig : spbetax1_t15.2 ] , [ fig : spbetax1_t30.4 ] , and [ fig : betax_fail ] make clear how the choice of the damping factor affects the behavior of the simulations .
the value we choose for @xmath1 should be non - zero and not larger than @xmath48 to allow for effective damping and stable simulations . the exact cutoff value between stable and unstable simulations is not relevant here since the position dependent form we develop in sec .
[ sec : forms_of_eta ] gives us the flexibility we need to obtain stable simulations . to examine the effect of @xmath1 on the extraction of gravitational waves , we compare the results from simulations of an equal mass binary with total mass @xmath9 in quasi - circular orbits with initial separation @xmath50 , using @xmath51 .
again , the courant factor is chosen to be @xmath26 and we use , in the terminology of @xcite , the grid configuration @xmath52 $ ] with a finest resolution of @xmath53 . here
, the extraction radius @xmath54 is chosen to be @xmath55 . for vanishing @xmath1
, we find a lot of noise in the the real part of the 22-mode of @xmath56 , shown in the solid curve of fig .
[ fig : emexx_rpsi4 ] .
a small , but non - vanishing @xmath1 suffices to suppress this noise , as seen in the dashed curve of this figure .
the dotted - dashed curve in this plot is the result for using the value @xmath57 .
we see a difference in time between peak amplitudes of the three curves due to the change of coordinates that the alternation of @xmath1 introduces .
we did , however , find that by decreasing the courant factor those differences between peak amplitudes summarily decreased .
to understand the noise in the waves for @xmath58 , we look at the shift vector at different times .
the first panel of fig .
[ fig : emexx_betax ] shows the @xmath31-component of the shift over @xmath31 , again for @xmath51 , shortly after the beginning of the simulation .
the fourth panel shows the same at a time shortly before the merger , and the two panels in the middle represent intermediate times .
we see clear gauge pulses in the earliest time panel for all three curves .
we also observe the amplitude of this pulse decreasing with increasing @xmath1 .
as time goes on , the gauge pulse travels outwards as in the case for a single puncture in section [ sec : single_hole ] . for vanishing @xmath1 ( solid line ) ,
the shift becomes more and more distorted , and the distortions do not leave the grid . for non - zero @xmath1 , the amplitude of the gauge pulse decreases when traveling outwards , and the shape of @xmath33 is not distorted .
there is , compared to @xmath57 ( dotted - dashed line ) , only a small bump left in the @xmath59 case ( dashed line ) , that changes its shape slightly during the simulation , but does not travel to large distances from the punctures .
the coordinates are disturbed in the case where no damping is used , and thus the noise in @xmath60 is not surprising . over time for equal mass simulations using different values for @xmath1 .
the inset shows the full waveform until ringdown .
the solid curves ( black ) are for @xmath61 , the dashed curve ( red ) mark @xmath59 and the dotted - dashed curves ( green ) are for @xmath57 . without damping in the shift ,
the extracted waves are noisy at times when the amplitude is still small ( black , solid curve ) . ]
-component of the shift vector , @xmath33 , for three different choices of @xmath1 at four different times during the simulation .
the physical system is the same as in fig .
[ fig : emexx_rpsi4 ] .
the merger takes place at approximately @xmath62 . in the @xmath63 case ( black , solid curve ) , the shift vector is not damped and therefore , a pulse travels outwards and distorts the shift over the whole grid .
the amplitude of this pulse is considerably damped when using a non - vanishing @xmath1 and therefore the distortions are reduced .
for @xmath59 ( red , dashed curve ) , there are still small bumps traveling out which are reduced by using @xmath64 ( green , dotted - dashed curve ) . ] in this series , using a courant factor of 0.5 , we only obtained stable evolutions for @xmath65 which agrees with the limits found in section [ sec : single_hole ] .
if we chose the value of @xmath1 too large , the same kind of instability in the shift vector we found there develops in the equal mass case and the simulations fail .
the failure occurs relatively early , before @xmath66 of evolution time , whereas the stable runs lasted about @xmath67 , including merger and ringdown ( we stopped the runs after ringdown ) .
in section [ sec : single_hole ] , we saw that a sufficient level of damping is needed to limit gauge dynamics , and too much damping can lead to numerical instabilities . in section
[ sec : embinary ] , we saw the positive effect that sufficient damping has on the resultant waveform for equal mass binaries .
while we still need damping in the gamma driver in the unequal mass case , a constant value may not fulfill the requirements of limiting gauge dynamics and permitting stable evolutions .
rather , we need a definition for the damping that adjusts the value to the local mass - scale
. we will examine definitions , that naturally track the position , and mass of the individual black holes .
the choice of @xmath1 should provide a reasonable value both near the individual black holes , and at large distance from the binary .
we will start by examining some previous work , that has used non - constant forms of the damping parameter , and why it may be necessary to use other formulas .
we will then present the two new formulas for @xmath1 , which we designed for this work .
a position dependent damping was introduced some years ago by the authors of @xcite , and was later used in @xcite .
that formula reads @xmath68 with @xmath69 , @xmath70 being constants , and assuming @xmath71 ( @xmath72 is the distance to the @xmath12 puncture ) .
this formula was used to damp gauge dynamics while using excision for equal - mass head - on collisions .
it has since been found that using the moving puncture framework allows for constant damping in the approximately equal mass case .
we are looking for a formula which is suitable for the quasicircular inspiral of intermediate mass - ratio binaries .
previously @xcite , we used the formula @xmath73 , along the @xmath31-axis using eq .
( [ eq : etapsimm ] ) .
the simulated configuration is an equal mass binary with initial separation @xmath50 and orbits lying in the @xmath74-plane . shown
are four different times during the simulation . ] for determining a position dependent damping coefficient instead of using a constant @xmath1 . with @xmath75 taken to be a unitless constant
, it can be seen that eq .
( [ eq : etapsimm ] ) has units of inverse mass .
the dependence on the bssn variable , @xmath15 , naturally tracks the position , and mass of the black holes .
the application of eq .
( [ eq : etapsimm ] ) gave good values for the damping both at the punctures , and at the outer boundary , and was even found to somewhat decrease the grid - size of the larger black hole .
the latter point could have positive effects on how the individual black holes are resolved on the numerical grid .
it even had the additional effect of keeping the horizon shapes roughly circular , even close to merger - something that does nt hold in the constant @xmath1 case .
most importantly , the simulations remained stable , without significantly changing the gravitational waves .
the formula was later used successfully for the 10:1 mass - ratio in @xcite . despite all this , eq .
( [ eq : etapsimm ] ) provides reason for concern .
[ fig : oldeta_waves ] shows the form of @xmath1 using eq .
( [ eq : etapsimm ] ) for a non - spinning binary of equal mass in quasicircular orbits starting at a separation of @xmath50 at four different times in the simulation .
as can be seen , noise travels out from the origin as time progresses .
this leaves steady features on the form of @xmath1 which could spike to higher and lower values than the range determined in sec .
[ sec : single_hole ] .
additionally , these sharp features may lead to unpredictable coordinate drifts , and could , in some cases , affect the long - term stability of the simulation .
to illuminate the origin of the disturbances in @xmath76 , we looked at the development of @xmath76 in simulations of a single , non - spinning puncture , and a single , spinning puncture ( @xmath77 ) .
the result for the spinning case is plotted in fig .
[ fig : kbh_eta ] at two different times over the @xmath31-axis .
again , we see a pulse traveling outwards . only this time
, it does not leave much noise on the grid .
the fact that this pulse travels at a speed which is roughly @xmath78 ( in our geometric units where @xmath79 ) in both the spinning and non - spinning scenario indicates that it is related to the gauge modes traveling at speed @xmath80 in the asymptotic region where @xmath81 ( see @xcite and @xcite for a discussion of gauge speeds ) .
in contrast to gauge pulses in the lapse , @xmath82 , or shift vector , @xmath3 , the pulse in @xmath83 is amplified as it walks out .
we found the same result in the single puncture simulation without spin .
we believe the reason for this behavior is that as the distance to the puncture increases , the conformal factor , @xmath15 , gets closer to unity .
therefore , the denominator in eq .
( [ eq : etapsimm ] ) approaches zero , and the gauge disturbances in the derivatives of @xmath15 are magnified .
we further observed reflections at the refinement boundaries as this pulse passes through them .
this may explain the fluctuations in @xmath83 shown in fig .
[ fig : oldeta_waves ] .
while one could continue to fine - tune a formula dependent on the conformal factor to deal with these problems , we looked in a different direction to determine the form of the damping parameter . for a single spinning puncture sitting at @xmath84 using eq .
( [ eq : etapsimm ] ) after simulation time @xmath85 ( solid black line ) and @xmath86(dashed red line ) over @xmath87direction . ] since we always know the location of a puncture , and we know what its associated mass , we chose a form of damping that uses this local information throughout the domain . to address the demands and concerns discussed in section [ sec : motivation ] and [ sec : prev ] , we designed two position - dependent forms of @xmath1 .
the two forms we tested are @xmath88 and @xmath89 in eqs .
( [ eq : etas6 ] ) and ( [ eq : etas5 ] ) , @xmath90 and @xmath91 are required to be positive , unitless parameters which can be chosen to change the width of the functions .
the power @xmath92 is taken to be a positive integer which determines the fall - off rate .
the constants @xmath93 , @xmath94 , and @xmath95 are then chosen to provide the desired values of @xmath1 at the punctures , and at at infinity .
lastly , @xmath96 and @xmath97 are defined as @xmath98 , where @xmath99 is either one or two , and @xmath100 is the position of the @xmath99th black hole .
the definition of @xmath101 is chosen to naturally scale the fall - off to the separation of the black holes .
@xmath90 , @xmath91 , and @xmath92 can be chosen to change the overall fall - off .
our work focuses on the choice @xmath102 and @xmath103 .
following @xcite , we construct the damping factor to have units of inverse mass .
we choose @xmath104 , where @xmath105 is defined as the sum of the irreducible masses .
we then take @xmath106 .
it is then evident that both eqs .
( [ eq : etas6 ] ) and ( [ eq : etas5 ] ) will give a constant value of @xmath107 in the equal mass case .
we designed the two formulas for @xmath1 in order to test the value of using fundamentally different functions . in our simulations , we found little noticeable difference in the application of one compared to the other . in the absence of such a difference , it becomes more beneficial to use eq .
( [ eq : etas6 ] ) , as gaussians are computationally more expensive .
it should be pointed out that eq .
( [ eq : etas6 ] ) is very similar to eq .
( 13 ) suggested in @xcite , and we believe the following results are very similar to what would be found using that form for the damping . going into the present work , we have no ansatz which might suggest these forms of damping yield wave forms which are any better than the use of any previous form of @xmath1 .
however , as will be seen in the results sections , the waveforms we get from unequal mass binaries show noticeable improvement over the constant @xmath1 case .
for data analysis purposes , we are mainly interested in the properties of the emitted gravitational waves of the black hole binary systems under study . hence , it is important to check how the changes in the gauge alter the extracted waves . in the context of gravitational wave extraction
, @xmath16 is only first order invariant under coordinate transformations .
in addition , we have to chose an extraction radius @xmath54 for the computation of modes , which is also coordinate dependent .
although the last point can be partly addressed by extrapolation of @xmath108 , it is a priori not clear how much a change of coordinates affects the gravitational waves .
furthermore , a change of coordinates implies an effective change of the numerical resolution , and for practical purposes we have to ask how much waveforms differ at a given finite resolution .
the results in the following section refers to the use of eq .
( [ eq : etas6 ] ) .
we compare numerical simulations using three different grid configurations , which correspond to three different resolutions . in the terminology of @xcite ,
the grid set - ups are @xmath109 $ ] , @xmath110 $ ] , and @xmath111 $ ] , which corresponds to resolutions on the finest grids of @xmath112 ( @xmath113 ) , @xmath114 ( @xmath115 ) and @xmath116 ( @xmath117 ) , respectively .
when referring to results from different resolutions , we will from here on use the number of grid points on the finest grid , @xmath118 , to distinguish between them .
in this subsection , we use @xmath119 and @xmath103 in eq .
( [ eq : etas6 ] ) . as test system
we use an unequal mass black hole binary with mass ratio @xmath120 and an initial separation of @xmath121 without spins in quasi - circular orbits . for orientation ,
[ fig : um4amp ] shows the amplitude of the 22-mode , @xmath122 , computed with the standard gauge @xmath64 ( displayed as solid lines ) and with the new @xmath76 using eq .
( [ eq : etas6 ] ) ( displayed as non - solid lines ) .
the three different colors correspond to the three resolutions .
the inset shows a larger time range of the simulation , while the main plot concentrates on the time frame around merger .
the plot gives a course view of the closeness of the results we obtain with standard and new gauges . in fig .
[ fig : um4s6ampdevres2 ] , we plot the relative differences between the amplitudes at low and medium ( solid lines ) , and medium and high resolution ( non - solid lines ) obtained with @xmath64 ( light gray lines ) as well as @xmath76 ( eq . ( [ eq : etas6 ] ) ) ( black lines ) . here , we find the maximum error between the low and medium resolution of the series using @xmath64 amounts to about @xmath123 ( solid gray curve ) . between medium and high resolution ( dashed gray curve ) , we find a smaller relative error , but it still goes up to @xmath124 at the end of the simulation . employing eq .
( [ eq : etas6 ] ) , the maximum amplitude error between low and medium resolution ( solid black line ) is only about @xmath125 , and therefore even smaller than the error between medium and high resolution for the constant damping case . between medium and high resolution , the relative amplitude differences for eq .
( [ eq : etas6 ] ) are in general smaller than the ones between low and medium resolution , although the maximum error is comparable to it ( dot - dashed black line ) .
we repeat the previous analysis for the phase of the 22-mode , @xmath126 . again , we compare the errors between resolutions in a fixed gauge .
figure [ fig : um4phasedevres ] shows that the error between lowest and medium resolution using @xmath64 ( solid gray line ) grows up to about 0.31 radians . for the differences between medium and high resolution ( dashed line )
we find a maximal error of 0.2 radians for @xmath64 .
for @xmath76 following eq .
( [ eq : etas6 ] ) , the phase error between low and medium resolution is only about 0.19 radians ( solid black line ) and decreases to 0.1 radians between medium and high resolution ( dot - dashed line ) .
again , employing the position dependent form of @xmath1 , eq .
( [ eq : etas6 ] ) , the error between lowest and medium resolution is lower than the one we obtain for constant @xmath1 between medium and high resolution .
the results for amplitude and phase error suggest that we can achieve the same accuracy with less computational resources using a position - dependent @xmath76 . of a binary with mass ratio 4:1 and initial separation @xmath121 .
the different colors correspond to three different resolutions according to the grid setup described in the text .
the solid lines are results for @xmath64 , the dashed , dotted and dot - dashed ones are for @xmath76 ( eq .
( [ eq : etas6 ] ) ) .
the inset shows the simulation from shortly after the junk radiation passed , in the main plot we zoom into the region of highest amplitude ( near the merger ) .
, title="fig : " ] ( -187,20 ) of a binary with mass ratio 4:1 and initial separation @xmath121 .
the different colors correspond to three different resolutions according to the grid setup described in the text .
the solid lines are results for @xmath64 , the dashed , dotted and dot - dashed ones are for @xmath76 ( eq .
( [ eq : etas6 ] ) ) .
the inset shows the simulation from shortly after the junk radiation passed , in the main plot we zoom into the region of highest amplitude ( near the merger ) .
, title="fig : " ] ( -50,105 ) of a binary with mass ratio 4:1 and initial separation @xmath121 .
the different colors correspond to three different resolutions according to the grid setup described in the text .
the solid lines are results for @xmath64 , the dashed , dotted and dot - dashed ones are for @xmath76 ( eq .
( [ eq : etas6 ] ) ) .
the inset shows the simulation from shortly after the junk radiation passed , in the main plot we zoom into the region of highest amplitude ( near the merger ) .
, title="fig : " ] between resolutions @xmath113 and @xmath115 ( gray solid curve ) as well as @xmath115 and @xmath117 ( gray dashed curve ) when using @xmath64 . the same for @xmath76 ( eq . ( [ eq : etas6 ] ) ) between @xmath113 and @xmath115 ( black solid curve ) and @xmath115 and @xmath117 ( black dot - dashed curve ) .
the physical situation is the same as in fig .
[ fig : um4amp ] .
the maximum differences are above @xmath127 , comparing low and medium resolution of the constant @xmath1 simulations ( gray solid line ) . ,
title="fig : " ] ( -200,100 ) between resolutions @xmath113 and @xmath115 ( gray solid curve ) as well as @xmath115 and @xmath117 ( gray dashed curve ) when using @xmath64 . the same for @xmath76 ( eq . ( [ eq : etas6 ] ) ) between @xmath113 and @xmath115 ( black solid curve ) and @xmath115 and @xmath117 ( black dot - dashed curve ) .
the physical situation is the same as in fig .
[ fig : um4amp ] .
the maximum differences are above @xmath127 , comparing low and medium resolution of the constant @xmath1 simulations ( gray solid line ) . , title="fig : " ] ( solid gray line ) and @xmath76 ( eq . ( [ eq : etas6 ] ) ) ( solid black line ) as well as between medium and high resolution for @xmath64 ( dashed gray line ) and for @xmath76 ( eq . ( [ eq : etas6 ] ) ) ( dot - dashed black line ) .
the physical situation is the one of fig .
[ fig : um4amp ] .
, title="fig : " ] ( -215,105 ) ( solid gray line ) and @xmath76 ( eq . ( [ eq : etas6 ] ) ) ( solid black line ) as well as between medium and high resolution for @xmath64 ( dashed gray line ) and for @xmath76 ( eq .
( [ eq : etas6 ] ) ) ( dot - dashed black line ) .
the physical situation is the one of fig .
[ fig : um4amp ] .
, title="fig : " ] we repeated the analysis of sec . [ sec : resultss6 ] with the waveforms we obtain using eq .
( [ eq : etas5 ] ) ( with @xmath119 and @xmath103 ) .
we use the same initial conditions ( mass ratio @xmath128 , @xmath121 , no spins ) , and compare the amplitudes and phases of the @xmath129-mode of @xmath16 with the results of the @xmath64-runs .
the grid configurations remain the same .
the results are very similar to the ones we obtained in figs .
[ fig : um4s6ampdevres2 ] and [ fig : um4phasedevres ] , and we therefore do not show them here .
although eqs .
( [ eq : etas6 ] ) and ( [ eq : etas5 ] ) result in different shapes for @xmath76 , @xmath130 is very similar .
therefore , the comparison to @xmath64 naturally gives very similar results , too .
the phase differences between results from eqs .
( [ eq : etas6 ] ) and ( [ eq : etas5 ] ) at a given resolution are shown in fig .
[ fig : um4phasedev_s5_s6 ] . these are , with a maximum phase error of 0.004 radians , very small compared to the phase errors between resolutions , which , at minimum , are about 0.1 radian ( see fig . [
fig : um4phasedevres ] ) . fig .
[ fig : um4phasedevress5s6psim2 ] compares the phase error between low and medium ( solid lines ) , and medium and high resolution ( dotted - dashed and dashed line ) of eq .
( [ eq : etas5 ] ) ( gray ) to the ones of eq .
( [ eq : etas6 ] ) ( black ) . for comparison ,
the error between medium and high resolution is also plotted for eq .
( [ eq : etapsimm ] ) in this figure ( dotted line ) .
the plot indicates that the errors between resolutions are in good agreement for the different position dependent formulas of @xmath1 . ) and
( [ eq : etas5 ] ) in three different resolutions ( solid , dashed , dotted - dashed lines ) for mass ratio 4:1 , @xmath121 .
, title="fig : " ] ( -215,100 ) ) and eq .
( [ eq : etas5 ] ) in three different resolutions ( solid , dashed , dotted - dashed lines ) for mass ratio 4:1 , @xmath121 . ,
title="fig : " ] ) ( black lines ) or eq .
( [ eq : etas5 ] ) ( gray lines ) for mass ratio 4:1 , @xmath121 . for comparison
, we also show the phase difference obtained with eq .
( [ eq : etapsimm ] ) between medium and high resolution ( dotted line ) .
, title="fig : " ] ( -215,80 ) ) ( black lines ) or eq .
( [ eq : etas5 ] ) ( gray lines ) for mass ratio 4:1 , @xmath121 . for comparison
, we also show the phase difference obtained with eq .
( [ eq : etapsimm ] ) between medium and high resolution ( dotted line ) .
, title="fig : " ] -component of the shift vector in @xmath31-direction after @xmath131 of evolution of the system with mass ratio @xmath128 and @xmath50 .
the black , dot - dashed line refers to the use of a constant damping @xmath1 , while the black , solid line uses eq .
( [ eq : etapsimm ] ) .
the gray , dashed line is for the use of eq .
( [ eq : etas5 ] ) and the gray , dotted one for eq .
( [ eq : etas6 ] ) . except for the constant @xmath1 ( black , dot - dashed line ) ,
the results in this plot are indistinguishable . , title="fig : " ] ( -90,80)-component of the shift vector in @xmath31-direction after @xmath131 of evolution of the system with mass ratio @xmath128 and @xmath50 .
the black , dot - dashed line refers to the use of a constant damping @xmath1 , while the black , solid line uses eq .
( [ eq : etapsimm ] ) .
the gray , dashed line is for the use of eq .
( [ eq : etas5 ] ) and the gray , dotted one for eq .
( [ eq : etas6 ] ) . except for the constant @xmath1 ( black , dot - dashed line ) ,
the results in this plot are indistinguishable .
, title="fig : " ] in @xcite , we found an unusual behavior of the shift vector .
this is illustrated in fig .
[ fig : um4betax ] , where we plot the @xmath31-component of the shift , @xmath33 , in the @xmath31-direction after @xmath131 of evolution ( this means approximately @xmath132 after merger ) for all four versions of the damping constant we used for comparison in this paper before , and for the same binary configuration as the one used in secs .
[ sec : resultss6 ] and [ sec : resultss5 ] . like in @xcite
, we find that using eq .
( [ eq : etapsimm ] ) results in a shift which falls off to zero too slowly towards the outer boundary , and which develops a `` bump '' ( black , solid line ) , while the constant damping case ( black , dot - dashed line ) falls off to zero quickly .
employing eqs .
( [ eq : etas6 ] ) or ( [ eq : etas5 ] ) avoids this undesirable feature . after merger
, the shift falls off to zero when going away from the punctures as it does in the constant damping case ( gray dashed and dotted lines ) . using eq .
( [ eq : etas6 ] ) or ( [ eq : etas5 ] ) prevents unwanted coordinate drifts at the end of the simulations .
in this work , we examined the role that the damping factor , @xmath1 , plays in the evolution of the shift when using the gamma driver . in particular
, we examined the range of values allowed in various evolutions , and what effects showed up because of the value chosen .
we then designed a form of @xmath1 for the evolution of binary black holes which provides appropriate values both near the individual punctures and far away from them with a smooth transition in between . in sec .
[ sec : results ] , we directly examined the waveforms for the case using eq .
( [ eq : etas6 ] ) , where @xmath119 and @xmath103 . while the form of @xmath1 is predictable , and can be easily adjusted for stability , we also saw that the waveforms produced using this definition showed less deviation with increasing resolution than using a constant @xmath1 . when examining the waveforms produced using eq .
( [ eq : etas5 ] ) , we found similar results . in the absence of a noticeable difference in the quality of the waveforms , eq .
( [ eq : etas6 ] ) is computationally cheaper , and , as such , is our preferred definition for the damping . with initial separation @xmath121 .
the black , blue and red lines use @xmath76 , eq . ( [ eq : etas6 ] ) with varying values of the width parameter @xmath133 . the orange line ( dash - dot - dot ) uses the constant damping @xmath64 and the green ( dash - dot ) one refers to the result of @xcite with eq .
( [ eq : etapsimm ] ) . using eq .
( [ eq : etas6 ] ) , the coordinate areas can be varied with respect to each other depending on the choice of @xmath133
. a ratio of 1 means the black holes have the same size on the numerical grid . ]
we have already pointed out a certain freedom to pick parameters in eqs .
( [ eq : etas6 ] ) and ( [ eq : etas5 ] ) .
we did perform some experimentation along this line where we varied @xmath134 to see if we could get a useful effect of the coordinate size of the apparent horizons on the numerical grid . in @xcite
, it was noticed that the damping coefficient affects the coordinate location of the apparent horizon , and therefore the resolution of the black hole on the numerical grid .
[ fig : ah_coordarea ] plots the ratio of the grid - area of larger apparent horizon to the smaller apparent horizon as a function of time for @xmath133-values of @xmath135 , @xmath26 and for @xmath136 , all with @xmath103
. also plotted is the relative coordinate size for the same binaries using a constant @xmath1 in dashed , double - dotted line , and for using eq .
( [ eq : etapsimm ] ) in a blue dashed - dotted line .
all the evolutions show an immediate dip , and then increase in the grid - area ratio during the course of the evolution . while a very low ratio was found using eq .
( [ eq : etapsimm ] ) , the orange dotted line was later found for the choices of @xmath137 with @xmath138 and @xmath139 with eq .
( [ eq : etas6 ] ) . due to this freedom in the implementation of our explicit formula for the damping
, it may be possible to further reduce the relative grid size of the black holes .
this effect could be important in easing the computational difficulty of running a numerical simulation for unequal mass binaries . having a form of @xmath1 that leads to stable evolutions for any mass - ratio is an important step towards the numerical evolution of binary black holes in the intermediate mass - ratio .
we believe the form given in eq .
( [ eq : etas6 ] ) provides such a damping factor at a low computational cost , although the test results presented are limited to mass ratio @xmath128 .
we plan to examine larger mass ratios in future work .
the new method should allow binary simulations for mass ratio @xmath140 , or even @xmath141 .
it remains to be seen whether other issues than the gauge are now the limiting factor for simulations at large mass ratios .
it is a pleasure to thank zhoujian cao and erik schnetter for discussions .
this work was supported in part by dfg grant sfb / transregio 7 `` gravitational wave astronomy '' and the dlr ( deutsches zentrum fr luft und raumfahrt ) .
d. m. was additionally supported by the dfg research training group 1523 `` quantum and gravitational fields '' .
computations were performed on the hlrb2 at lrz munich . | certain numerical frameworks used for the evolution of binary black holes make use of a gamma driver , which includes a damping factor .
such simulations typically use a constant value for damping .
however , it has been found that very specific values of the damping factor are needed for the calculation of unequal mass binaries .
we examine carefully the role this damping plays , and provide two explicit , non - constant forms for the damping to be used with mass - ratios further from one .
our analysis of the resultant waveforms compares well against the constant damping case . |
active galactic nuclei ( agn ) feature narrow - line regions ( nlrs ) in which continuum radiation from the central engine photoionizes the surrounding interstellar medium ( ism ) .
the relatively hard agn continuum emission injects more energy and photoionizes the ism to higher ionization states than the ionizing spectra of ob stars at the centers of regions .
many insights into the physics of the the ism in active galaxies have been gained through the use of photoionization models .
the results of photoionization modeling facilitate the routine diagnosis of the density , temperature , and cause of ionization of nlr gas from observed emission - line spectra .
photoionization models have been used to explain the very similar nlr ionization parameters observed across the population of active galaxies - this effect is caused by the gas pressure gradient balancing the radiation pressure gradient in dusty nlr clouds such that the ` local ' ionization parameter is self - regulated @xcite .
models are an important tool in investigating the effects of various ism parameters on emission - line spectra ; e.g. determining the effects of changing abundances @xcite .
recently , @xcite showed how comparisons between models and spatially - resolved spectroscopy may reveal where extended nlrs are radiation pressure - dominated or gas pressure - dominated on scales of tens to hundreds of parsecs .
the emission line spectra of nlrs in active galaxies strongly depend upon the shape of the spectrum of the central source at energies immediately above the lyman limit at 13.6 ev .
these ionizing photons are not directly observable because they are very strongly absorbed in the nlr and broader ism of the host galaxy , as well as in the ism of the milky way . up to the present ,
ionizing agn spectra for photoionization modeling have almost exclusively been constructed using piecewise functions of one to three simple power - laws ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
@xcite use a conventional range of power - law indices of @xmath3 to @xmath4 , for a single power - law covering 0.005 to 1 kev .
the authors find that a steeper power - law leads to weaker high - ionization lines and relatively stronger h lines , due to the relative excess of h- but not he - ionizing photons .
a more physically - motivated empirical two - component model for the ionizing spectrum is used , for example , in @xcite ) . in this work
we seek to develop a modern agn continuum emission model for use as the input spectrum in photoionization models , based on physically - derived models , and including for example ionization effects in the shape of the accretion disk spectrum .
we aim to identify the important agn parameters which influence the shape of the ionizing spectrum and hence the predicted emission - line ratios .
this allows us to minimize the parameter space used to cover a large range of model spectra .
a small parameter space is easier to handle conceptually and computationally , and the process of combining fully and partially degenerate parameters into a smaller set of parameters results in insights into how the shape of the ionizing spectrum may arise from various combinations of fundamental agn parameters . in photoionization modeling
we assume that the model ionizing spectrum will be intrinsic when it reaches the modeled clouds , i.e. that there is no significant absorption between the central engine and the modeled clouds .
hence the spectra will primarily feature a peak formed by a superposition of pseudo - blackbodies due to the accretion disk , and a power - law non - thermal component resulting from compton up - scattering of disk photons . a physical model of these spectra
may be expected to be well - described by few parameters .
a detailed model of continuum emission from agn must rely on theoretical emissivities from accretion disk models .
the standard novikov - thorne ` thin disk ' accretion model ( @xcite , building on the work of @xcite ) is applicable over much of the accretion rate regime of seyfert galaxies ( @xmath5 to @xmath6 @xmath7 ) .
this model has parameters such as the smbh mass , spin , and accretion rate , which influence the theoretical ionizing agn spectrum . understanding how the key parameters control the shape of the ionizing euv - soft x - ray spectrum of agn allows us to produce a model which keeps the number of free parameters to a minimum .
our simplified spectral modeling is , however , based firmly on the optxagnf model @xcite .
we chose this model because it is a recent , widely - used model , it calculates theoretical spectra using physics such as the novikov - thorne thin accretion disk model , and because it uses a ` color temperature correction ' which accounts for electron - scattering opacity in parts of the disk where hydrogen is ionized .
the optxagnf model provides the basis for a new physically - based model of the ionizing agn spectrum for photoionization modeling .
in photoionization modeling , the ionization parameter and density are important input parameters , so the flux incident on the model cloud is easily calculated .
however the actual luminosity of the ionizing source is important only if the modeler is interested in determining the distance of the modeled cloud from the source .
we are primarily concerned with the shape of the ionizing spectrum , because this determines the relative emission - line fluxes .
any parameter redundancies may render more difficult the interpretation of the nlr spectrum caused by a given agn ionizing spectrum . in this work
we identify and explain various spectral - shape degeneracies , such as the complete degeneracy between the black hole mass and luminosity in the novikov - thorne emissivity , and successfully remove the degeneracies by constructing a new model .
we simplify the optxagnf model by carefully removing and merging degenerate parameters . the resulting model , oxaf ,
satisfies our requirements by generating realistic model spectral shapes using the fewest possible parameters , and with all of the parameters having a direct impact on the shape of the model spectrum .
we then investigate the effect of the oxaf parameters on photoionization models of nlr clouds . in section [ sec : accretion_models ] we describe the spectral - shape degeneracies that arise in the thin - disc accretion model and in the the optxagnf model , and in section [ sec : param_reduction ] we describe the parameters used to re - parametrize optxagnf . in section [ sec : oxaf ]
we explain how these parameters are used to construct our simplified model oxaf and describe the implementation and validation of the new model . in section [ sec : oxaf_predictions ] we explore how the oxaf parameters affect photoionization model predictions for diagnostic emission lines .
our conclusions are presented in section [ sec : conclusion ] .
the unified model of agn describes a central , accreting supermassive black hole ( smbh ) surrounded by a broad line region , a dusty obscuring torus , and a narrow - line region @xcite .
the powerful continuum radiation from the central accretion structure illuminates the broad and narrow - line regions .
important parameters which determine the spectrum of the central accretion structure are the eddington ratio @xmath7 , the smbh mass @xmath8 , the dimensionless smbh spin @xmath9 , and the innermost radius at which the disk is directly visible . in this section
we describe the effects of these parameters on the shape of the ionizing spectrum and the relationships between the parameters .
the novikov - thorne thin disk model produces a clean relationship between @xmath8 and @xmath10 , with these parameters having the same effect on the model spectral shape .
the temperature @xmath11 at a given radius of the accretion disk @xmath12 may be calculated using the following equation for the ` outer ' region of the disk ( equation 5.10.1 in * ? ? ?
* ) : @xmath13^{\frac{1}{4 } } \label{eq : t1}\ ] ] where @xmath8 is the black hole mass , @xmath14 is the accretion rate , @xmath9 is the dimensionless spin parameter , @xmath15 is the stefan - boltzmann constant , and @xmath16 is the gravitational constant . with @xmath17 defined as the radius normalized by the gravitational radius @xmath18 ( @xmath19 ;
@xmath17 = @xmath20 ) , the quantities @xmath21 , @xmath22 and @xmath23 are complicated dimensionless radial functions associated with relativistic corrections in the analytic accretion disk solution of @xcite ( the @xmath24 function was presented in the follow - up work of @xcite ; @xmath25 , @xmath26 and @xmath24 tend to 1 at large @xmath17 ) . using @xmath27 where @xmath28 is the total accretion efficiency , and with the definition of @xmath17 given above , equation [ eq : t1 ]
may be rewritten as @xmath29^{\frac{1}{4 } } \label{eq : t2}\ ] ] where @xmath30 is a collection of physical and numerical constants and @xmath31 .
inverting equation [ eq : t2 ] , we find that @xmath32 for a function @xmath33 , since @xmath34 .
specifying the temperature at a particular normalized radius as well as the spin uniquely determines the ratio @xmath35 such that the parameters @xmath7 and @xmath8 are degenerate in their effects on the spectral shape .
the optxagnf model @xcite is a significant effort in the development of theoretical models of continuum radiation from accretion onto black holes that match observed optical , uv and x - ray continuum spectra .
the model is available for use with the xspec spectral fitting package , under the name optxagnf .
the model aims to capture the essential features of the continuum emission from a relativistic thin accretion disk with a comptonizing corona in the equatorial plane of a rotating black hole .
it is assumed that material is accreted only through the outer disk , and the released gravitational energy is divided between three spectral components : * a pseudo - thermal accretion disk component * a high - energy , power - law non - thermal component , formed by compton up - scattering from an optically thin , high - temperature medium * an intermediate ( ` soft x - ray excess ' ) component formed by compton up - scattering from a compton - thick , low - temperature medium the model implements the following : * the standard @xcite thin - disk relativistic accretion disk emissivity * a disk spectrum produced by summing blackbodies in successive annuli , with the temperature of the blackbody spectrum corrected if necessary using an empirical color temperature correction @xmath36 .
the correction is required due to the absorption opacity varying with temperature , density and wavelength ; in particular it is used where hydrogen is ionized in the inner parts of the disk in an attempt to account for modifications to the spectrum due to the disk not being locally thermalized and due to electron - scattering opacity @xcite .
the correction becomes increasingly important as the temperature of the disk increases ( as the luminosity increases or black hole mass decreases ) . * the disk luminosity between the innermost stable circular orbit @xmath37 and @xmath38 , the ` coronal radius ' ,
is all emitted as comptonized radiation . *
the comptonized radiation is split between the soft , intermediate component ( assumed to arise in the accretion structure itself ) and the hard x - ray power - law tail .
the seed photons for comptonization in both cases come from a blackbody with the same ( corrected ) temperature as the disk at @xmath38 .
* the total luminosity from all three components of the emission ( disk , comptonized soft x - ray excess , comptonized hard x - ray tail ) is fixed by the accretion rate and smbh spin .
the following are the parameters required by optxagnf to characterize the various physical components of the model : * @xmath8 , the black hole mass * @xmath10 , the agn luminosity in units of the eddington luminosity , which is proportional to @xmath8 .
* @xmath9 , the dimensionless black hole rotation parameter , between 0 and 1 ( assuming prograde accretion ) * @xmath38 , the coronal radius ( inner edge of the visible portion of the disk ) , in units of the gravitational radius ( @xmath19 , proportional to @xmath8 ) * @xmath39 , the outer radius of the modeled disk , in units of the gravitational radius , proportional to @xmath8 * @xmath11 , the temperature of the compton optically thick material which produces the soft x - ray excess * @xmath40 , the optical thickness of the compton optically thick material * @xmath41 , the proportion of the corona power emitted in the hard power - law component ; @xmath42 is the proportion in the compton optically thick component * @xmath1 , the negative of the photon power - law index of the hard x - ray tail component we explored the behavior of the disk and non - thermal power - law components of the optxagnf model across a wide parameter space ( the parameter ranges given in section [ sec : reparametrizing ] ) and showed that there are strong spectral - shape degeneracies between key parameters .
the following sections discuss these degeneracies , which are in addition to the total degeneracy between @xmath8 and @xmath10 which originates in the novikov - thorne model .
higher disk temperatures lead to higher - energy disk emission , and in particular the high - energy cutoff of the disk emission is set by the maximum temperature of the directly visible part of the accretion disk .
increasing the temperature of the highest - temperature visible parts of the disk and thereby producing higher - energy disk emission is achievable by increasing @xmath35 ( increasing the luminosity and/or decreasing the mass ) , or alternatively by reducing @xmath38 .
consequently @xmath38 is partially degenerate in its effects on the disk spectrum with @xmath7 and @xmath8 .
the degeneracy is not total , because changing the innermost visible normalized radius modifies the shape of the disk spectrum in ways that changing @xmath35 can not . changing the spin @xmath9 with the other parameters fixed has the effect of changing the relative flux contribution of the disk and non - thermal components .
this effect is illustrated in figure [ fig : a_vary ] , which shows models with no intermediate component .
as the spin of a smbh increases , more energy is released by infalling matter ; the additional energy is released close to the smbh event horizon . when we fix the mass , coronal radius and luminosity , increasing the spin serves primarily to apportion more of the fixed luminosity to the part of the disk within the coronal radius , and hence to the non - thermal component .
hence @xmath9 has a similar effect to @xmath38 in that it changes the relative contributions of the disk and non - thermal components .
figure [ fig : a_vary ] shows that the shape of the two spectral components is not strongly affected by @xmath9 ( with @xmath43 @xmath44 .
however the parameters @xmath38 and @xmath9 are not entirely degenerate in their effects on the spectral shape , since @xmath9 affects the spectral shape through modifying the radial functions @xmath21 , @xmath22 and @xmath23 in equation [ eq : t1 ] , whereas @xmath38 determines which values of @xmath17 contribute to the disk emission .
the full and partial degeneracies discussed in the previous sections suggest that @xmath8 , @xmath7 , @xmath38 and @xmath9 could be combined into two parameters in a simplified model of the spectral shape - one parameter which shifts the disk emission in energy space , and one parameter which controls the relative flux between the disk and non - thermal components . , with the other key optxagnf parameters fixed at @xmath45 , @xmath46 , @xmath47 , and @xmath48 . the intermediate component is not considered ( @xmath49 ) .
varying @xmath9 changes the relative contributions of the disk and non - thermal components to the total spectrum .
note that the shape of these components remains relatively independent of @xmath9.,scaledwidth=49.0% ]
we reduce the free parameter space of the optxagnf model by fixing or neglecting some parameters , and by reparametrizing the remaining parameter space .
the origin of the widely - observed soft x - ray excess in agn remains uncertain , despite recent progress .
in particular , two general explanations are currently favored , which attempt to explain the relatively constant energy of the soft excess ( @xmath50 @xmath51 kev ) .
these explanations are comptonized disk emission ( e.g. the treatment in optxagnf ) , and blurred reflection from the ionized accretion disk ( e.g. * ? ? ?
additionally , absorption has been seriously considered ( e.g. * ? ? ?
* ) , and there is evidence that a contribution to the soft x - ray excess comes from the wider nlr ( e.g. * ? ? ?
the possible physical processes that give rise to the centrally - originating soft excess may be constrained using hard and soft time lags ( hard x - rays lagging soft x - rays and soft x - rays lagging hard x - rays ) and correlations between spectral components .
the time - delay of soft x - ray time lags is strongly correlated with black hole mass @xcite .
each soft x - ray excess photon with energy in the range @xmath52 kev will have less effect on the ionization state of the nebula than a photon with energy in the range @xmath53 kev . in the pilot study of @xcite we found that the intermediate soft x - ray component was not required to achieve satisfactory fits to the optical emission - line spectrum of the seyfert galaxy ngc 5427 with photoionization models , although the observed spectrum did not include high - ionization lines .
we do however expect that the abundances of high ionization potential species such as ( 423 ev ) , ( 361 ev ) and ( 234 ev ) would be affected by dominant soft x - ray excesses .
additionally , a strong soft x - ray excess extends the partially - ionized zone , where auger electrons produced by x - ray bombardment partially ionize the gas , and therefore must to some extent enhance emission of lines such as [ ] and [ ] .
we performed an investigation on the effect of the soft excess on photoionization modeling of nlr clouds , using ionizing spectra from the optxagnf model .
the parameters of the mappings 5.1 ( sutherland et al .
2016 , in prep . )
models and optxangf spectra were fixed , apart from two quantities : the relative strength of the soft excess which ranged from @xmath54 to @xmath55 of the non - thermal flux , and the coronal radius which ranged from @xmath56 to @xmath57 .
the models were ionization - bounded and ended when the nebula was 99% neutral .
the results are presented in standard optical diagnostic diagrams @xcite in figure [ fig : bptvo_soft_excess ] , with the fiducial model parameter values listed in the caption .
the standard optical diagnostic diagrams use ratios of lines associated with different ionization levels in the gas to distinguish nebulae based on the cause of ionization .
ionization increases from bottom - left to top - right on the diagrams .
the lower - left of each of the diagnostic diagrams is associated with regions of varying metallicity and ionization parameter , whereas the top - right is associated with ionization by harder agn spectra or shocks .
the standard dividing lines on the diagrams are described in the caption .
the results shown in figure [ fig : bptvo_soft_excess ] confirm that the soft x - ray excess is not of primary importance when predicting diagnostic optical emission - line ratios .
the line ratios are evidently most sensitive to variation in the coronal radius or weighting of spectral components when most of the energy released inside the coronal radius is emitted in the power - law component as opposed to the soft excess .
for the tested @xmath58 values larger than 20 @xmath18 , the [ ] /h@xmath59 and [ ] /h@xmath59 ratios are significantly increased by increasing the proportion of the flux emitted in the power - law tail , resulting in some outlying points to the right of the second and third panel in the figure .
the positions of these points are due to the partially - ionized zone being extended by hard x - rays in improbably hard ionizing spectra .
if we do not consider these power - law dominated models ( shown for completeness ) , the range of points representing the different soft excess strengths is small compared to the variation that may be produced by varying the ionization parameter or metallicity , for example .
if the proportion of the energy emitted in the soft excess as opposed to the power - law component does not vary as strongly between real objects as it does in our experiment , and additionally if real objects are mass - bounded as opposed to ionization - bounded ( i.e. hard x - rays escape the nlr ) , then the ranges shown in figure [ fig : bptvo_soft_excess ] would represent approximate upper bounds on the sensitivity to the soft excess . in the remainder of this work
we do not include the intermediate comptonized component , such that the fraction of the corona power in the power - law tail is set to @xmath60 , and the parameters for the temperature and optical thickness of the comptonizing optically - thick medium have no effect . neglecting the intermediate component
enables a vast simplification of the model . including the soft excess would complicate the comparison of photoionization models to observed nlr emission - line spectra due to uncertainty regarding the origin , time - averaged behavior , and importance relative to other spectral components of the soft x - ray excess .
we note that recently @xcite attributed @xmath61 of the @xmath62 kev emission observed in the agn ii zw 177 to the soft excess , with significant variation observed between observations separated by 11 years .
another recent work is a study of the agn re j1034 + 396 @xcite , in which the best - fitting models suggested that @xmath63 of the power produced inside the coronal radius is emitted in the intermediate component . although the strength of the soft excess has been measured in sources such as ii zw 177 and re j1034 + 396 , the properties of the soft x - ray excess across the population of seyfert galaxies , not just those that have been studied in x - rays , are unknown .
the spins of astrophysical black holes have long been expected to be non - zero .
it is now routine to ` map ' the central @xmath64 @xmath18 of agn accretion structures by studying x - ray reverberation lags and blurred accretion disk reflection @xcite .
identifying the innermost radius required by spectral blurring analysis with the innermost stable circular orbit ( isco ) allows the black hole spin to be measured .
spin measurements performed to date generally demonstrate high spins @xcite . however because radiative efficiency increases by up to a factor of 5 with increasing spin , the very high calculated spins for many agn with spin measurements are associated with a strong selection bias @xcite .
as demonstrated above in section [ sec : optxagnf_degeneracies ] , the black hole spin is partially degenerate with @xmath38 in that it changes the relative contributions of the disk and non - thermal components to the total spectrum .
the coronal radius will be set larger than radii for which we expect the disk spectrum shape to be strongly affected by the smbh spin .
figure [ fig : a_vary ] shows that changing the spin changes the relative energy in the disk and non - thermal components by an order of magnitude ( between spins of 0 and 1 ) ; in our simplified model this effect will be accounted for by an explicit parameter for this scaling .
the smbh spin was set to @xmath65 .
the outer disk produces non - ionizing optical and ir emission , and therefore does not contribute to the nlr heating .
the parameter defining the outer edge of the accretion disk was set to @xmath66 @xmath18 , a default value . as a result of the simplifications discussed in the section [ sec : neg_params ] ,
the optxagnf parameters have been reduced to the four most important parameters @xmath8 , @xmath67 , @xmath38 and @xmath1 .
exploration of the properties of the theoretical agn spectra and development of a simplified model ( section [ sec : oxaf ] ) were performed over the following ranges of these four parameters : * @xmath68 * @xmath69 * @xmath70 * @xmath71 @xmath72 @xmath18 the maximum luminosity of eddington was chosen because the novikov - thorne thin disk model can not plausibly be used beyond this limit ; indeed a _
slim _ disk prescription is necessary for a luminosity above @xmath6 @xmath7 .
the range of smbh masses covers the observed range in agn except for extreme objects ( c.f .
figure 4 in @xcite ) .
the broad range in @xmath1 covers the approximate range of observed values ( c.f .
figure 4 in @xcite ) , corresponding to very soft ( @xmath1 = 2.6 ) through to very hard ( @xmath1 = 1.4 ) power - law tails .
the physical extent of the corona has been constrained using analysis of blurred x - ray reflection , modeling of x - ray reverberation , and microlensing observations ; these methods suggest that the comptonizing corona lies within approximately @xmath71 @xmath18 of the black hole @xcite .
however because of the selection bias discussed in section [ sec : bh_spin_neglected ] , coronal geometries inferred in x - ray studies are not necessarily representative of the population of seyfert galaxies .
we chose a lower bound for the coronal radius of 10 @xmath18 here ; our experimentation showed that below this value the high - energy part of the accretion disk emission becomes too strongly affected by relativistic corrections for easy simplification of the model .
setting a minimum coronal radius of 10 @xmath18 will primarily affect the high - energy part of the accretion disk emission by removing it from the disk spectrum .
the top end of the @xmath38 range was selected to include the range suggested by @xcite .
a highly desirable simplification is combining the three fully or partially degenerate parameters @xmath67 , @xmath8 and @xmath38 into a single parameter that is able to parametrize ( at least approximately ) the position in energy space and shape of the big blue bump ( bbb ) disk emission .
the parameter which was chosen to replace the degenerate trio ( @xmath67 , @xmath8 , @xmath38 ) was @xmath0 , the peak of the bbb when the disk spectrum is given in a log - log plot of energy flux versus energy ( e.g. figure [ fig : a_vary ] ) .
the location of the single peak is a reliable and easily - calculated feature of a model bbb .
the value of @xmath0 was determined for a grid of optxagnf spectra over the relevant ranges of the three parameters affecting @xmath0 , which are @xmath67 , @xmath8 and @xmath38 .
the results were systematically analyzed for each value of @xmath38 , before all the analyses for various @xmath38 values were combined into an empirical formula for calculating @xmath0 as a function of ( @xmath67 , @xmath8 , @xmath38 ) .
an example of the analysis that was applied for a single value of @xmath38 is presented in figure [ fig : e_peak_pred_r24 ] . .
here we show the @xmath73 predicted by optxagnf as a function of @xmath74 ( blue points ) and @xmath8 ( blue curves as labeled , spaced at 0.5 dex ) , for @xmath75 @xmath18 .
to predict @xmath73 for a given @xmath58 and @xmath8 we use a model with the functional form @xmath76 ( equation [ eq_e_peak ] ) where @xmath77 is a power - law ( at low @xmath74 ) and @xmath78 is a log - space cubic ( at high @xmath74 ) .
both @xmath77 and @xmath78 are illustrated for @xmath79 by the red dashed curves .
model fits for @xmath80 and @xmath81 @xmath82 are overlaid in black .
, scaledwidth=49.0% ] figure [ fig : e_peak_pred_r24 ] demonstrates how fits for @xmath0 at a given @xmath38 must account for the ` piecewise ' behavior of @xmath0 as a function of @xmath8 and @xmath67 .
our solution to this problem was to use a combination of linear and cubic functions as described in the figure and caption .
the coefficients of these linear and cubic functions varied as a function of @xmath38 .
to construct a general function applicable over the entire three - dimensional ( @xmath67 , @xmath8 , @xmath38 ) parameter space , we required a means to determine the coefficients of the linear and cubic functions from @xmath38 .
linear and quadratic functions of @xmath83 produced satisfactory fits to the values of these coefficients .
the analysis in figure [ fig : e_peak_pred_r24 ] was applied for @xmath84 , 10 , 15 , 24 , 39 , 63 , 100 @xmath18 . the analysis for @xmath84 @xmath18 ( i.e. at the innermost stable circular orbit , such that the corona is effectively absent )
was found to be generally inconsistent with the analyses for the larger @xmath38 values , presumably due to the effect of strong general relativistic corrections .
we thus did not use the analysis for @xmath84 @xmath18 .
the combined analysis for the various @xmath38 values culminated in the following equation for a hypersurface to predict @xmath0 as a function of @xmath67 , @xmath8 and @xmath38 : @xmath85 where @xmath86 and @xmath87 here the energy of the bbb peak @xmath0 is in kev , @xmath88 where @xmath8 is in m@xmath89 , and @xmath90 .
the fit parameters are given in table [ tab : e_peak_params ] .
note that @xmath67 and @xmath8 appear explicitly in equations [ eq : f1 ] and [ eq : f2 ] only through the quantity @xmath91 , due to the complete degeneracy in their effects on the spectral shape ( section [ sec : m_l_degeneracy ] ; note that there is an implicit dependence on @xmath8 in the normalization of @xmath67 and @xmath38 ) .
the arbitrary constant 6 occurs with @xmath91 in equation [ eq : f2 ] only because it was propagated through the analysis from the cubic functions such as those in figure [ fig : e_peak_pred_r24 ] , where the cubic was fitted to the @xmath92 data ( the lowest @xmath8 value we considered ) , and was shifted appropriately to apply to other @xmath8 values .
equations [ eq_e_peak ] to [ eq : f2 ] in conjunction with table [ tab : e_peak_params ] predict @xmath0 with a standard deviation of @xmath93 in linear energy space .
the shape of the bbb spectrum depends ( to first order ) only on the energies it covers , so we use @xmath0 to parametrize both the location in energy space and shape of the bbb emission .
the construction of a simplified bbb model using this approach is discussed in section [ sec : bbb_modeling ] .
rrrrrrrr @xmath94 & -0.18 @xmath95 0.04 & & @xmath96 & 0.391 @xmath95 0.008 + @xmath97 & 0.0 @xmath95 0.1 & & @xmath98 & -0.83 @xmath95 0.02 + @xmath99 & 0.59 @xmath95 0.08 & & @xmath100 & 0.61 @xmath95 0.02 + @xmath101 & 0.250 @xmath95 0.004 & & @xmath102 & -0.30 @xmath95 0.03 + @xmath103 & 0.034 @xmath95 0.002 & & @xmath104 & 0.4 @xmath95 0.1 + @xmath105 & -0.019 @xmath95 0.003 & & @xmath106 & -0.82 @xmath95 0.07 the shape of the non - thermal component is determined by the shape of the seed photon spectrum and by the parameter @xmath1 .
the seed photon spectrum is a blackbody with a temperature which will be determined by the single parameter @xmath0 , so we have already reparametrized the shape of the non - thermal component such that it depends on only two parameters .
the construction of the non - thermal emission component is discussed in section [ sec : nt_modeling ] .
this section describes the construction of the oxaf model .
in particular oxaf is a 3-parameter model , and the name oxaf , a contraction of optxagnf , was chosen to reflect that oxaf is in some sense a ` reduced ' optxagnf .
the three parameters of the oxaf model are the energy of the peak of the big blue bump ( bbb ) @xmath0 , the photon power - law index @xmath1 , and the proportion of the total energy flux emitted in the non - thermal component @xmath107 , with the remainder being in the bbb .
in optxagnf the shape of the bbb and the proportion of energy which goes to the power - law component are not independent , because both are affected by @xmath38 .
however in oxaf , the parameters @xmath0 ( which determines the bbb shape ) and @xmath107 are independent by construction . for a given @xmath0
, oxaf does not have any parameters available to reproduce the variation in optxagnf bbb shapes which occurs as @xmath38 changes .
however the variation of the bbb shape with @xmath38 is a small effect , so a single oxaf bbb is able to satisfactorily reproduce optxagnf bbbs which have the same peak energy but a range of @xmath38 ( section [ sec : comparison ] ) .
the following sections show how we construct a bbb using only one parameter , the energy of the bbb peak @xmath0 , and a non - thermal component using only two parameters ( @xmath0 and @xmath1 ) .
the bbb is parametrized using only one parameter , the energy of the bbb peak @xmath0 .
the oxaf model must use this input to determine the shape of the bbb , in particular for relatively high bbb peak energies , when the color correction becomes important .
the modeled bbbs are considered in terms of log - log plots of energy flux ( i.e. kev/@xmath108/s ) versus energy .
the effects of the color correction in this space are illustrated in figure [ fig : bbb_shape ] . and
for @xmath109 @xmath18 .
the models were generated using a standalone version of optxagnf produced using a simple fortran wrapper .
the shape of the bbb changes due to the increasingly important color correction for higher - energy bbbs .
the curves are normalized to have the same flux in the bin with maximum flux , and shifted in logarithmic energy space to have coinciding peaks.,scaledwidth=47.0% ] the oxaf model constructs the bbb as follows : sixth - order polynomials were fit to a series of optxangf bbbs at different energies ( @xmath110 = @xmath111 , @xmath112 , @xmath113 , @xmath114 , @xmath115 , @xmath116 ) roughly representative of the range of shapes shown in figure [ fig : bbb_shape ] .
the bbbs were normalized to have the same peak flux and position of the peak flux in energy space before fitting .
the coefficients of the fits are stored in the oxaf code . to determine the shape of an oxaf bbb for a given input @xmath0 , the code ( log-)linearly interpolates between the polynomial fits for the nearest @xmath0 values above and below the input @xmath0 .
the resulting polynomial bbb shape is a weighted sum of the two ` nearest ' polynomials .
the interpolated polynomial is shifted in energy space so the peak aligns with the input @xmath0 value .
the non - thermal component is parametrized using two quantities : @xmath0 and @xmath1 .
the oxaf model uses the _ nthcomp _ routine to generate an inverse - compton scattered non - thermal component .
the fortran subroutines _
@xcite , and _ thcompton _ with _ thermlc _
@xcite from xspec were incorporated into oxaf .
there is one difference in the modeling of the non - thermal component between oxaf and optxagnf . in optxagnf
the seed blackbody spectrum temperature is set to @xmath117 , the color - corrected temperature of the disc at @xmath118 . to approximate this method in oxaf
it was necessary to convert the @xmath0 parameter to an estimate of @xmath117 , despite the oxaf model lacking an explicitly calculated @xmath118 .
the conversion was achieved by extracting @xmath117 values for 72 representative optxagnf models across the parameter space and relating them to the @xmath0 values calculated for these models using equation [ eq_e_peak ] .
the following linear fit from this comparison is used by oxaf to relate the two parameters : @xmath119 where both @xmath120 and @xmath0 are measured in kev . here the gradient and intercept have errors of 0.02 and 0.03 dex respectively , and the resulting error on @xmath121 is 0.07 dex .
the non - thermal spectrum is constructed by converting @xmath0 to @xmath117 using equation [ eq : t_r_cor ] before inputting @xmath1 , @xmath117 and an assumed comptonizing plasma temperature of 100 kev into the oxaf _ donthcomp _ function , which in turn calls _
thcompton _ , which calls _
thermlc_. the oxaf model was implemented as a module _ oxaf.py _ written in the programming language python , with a design focus on convenience of use .
the module depends only on the standard third - party numerical python library _ numpy _ , is a single file which may be used with both python 2 and python 3 , and requires no compilation step to run ( the included fortran codes were converted to python ) .
the module may simply be run as a command line script to output a model spectrum to stdout , or alternatively may be imported to be used in python code .
the module is thoroughly self - documented and is available on the astrophysics source code library ( ascl ) .
the _ oxaf.py _
module includes functions to find @xmath0 , the peak of the big blue bump ( bbb ) disk emission ( i.e. implement equation [ eq_e_peak ] ) , calculate an accretion disk spectrum ( as described in section [ sec : bbb_modeling ] ) , calculate a non - thermal component spectrum ( as described in section [ sec : nt_modeling ] ) , and sum these two components using a given weighting .
the output spectrum is a function of only three parameters , being @xmath110 , @xmath1 , and @xmath107 , which is the proportion of the total flux over the range @xmath122 ( kev ) @xmath123 which is assigned to the non - thermal component , with @xmath124 being the proportion assigned to the bbb disk component . the functions which return the individual components and the full spectrum ( sum of the two components ) normalize the spectra so that the sum of the bin fluxes over the range @xmath125 is equal to one .
the oxaf model was validated against the optxagnf model by comparing oxaf output with optxagnf output for many models across the parameter space .
the bbb and non - thermal components were considered separately .
the optxagnf parameters used were chosen so that their various combinations did not give multiple identical bbb peak energies and hence multiple identical oxaf spectra .
the chosen values were @xmath126 = @xmath127 , -3.75 , -2.85 , -2.15 , -1.60 , -1.20 , -0.80 , -0.40 , 0.00 ; @xmath128 = @xmath129 , 6.5 , 7.0 , 7.5 , 8.0 , 8.5 , 9.0 ; @xmath38 = @xmath71 , 25 , 40 , 50 , 63 , 100 @xmath18 and @xmath1 = 1.4 , 1.7 , 2.0 , 2.3 , 2.6 , for a total of 1890 optxagnf spectra .
the range of parameters was chosen for completeness ; not all of the tested parameter space necessarily corresponds to observed agn .
figures [ fig : oxaf_validation_bbb ] and [ fig : oxaf_validation_nt ] in the appendix demonstrate the performance of the oxaf model in reproducing the output spectra of optxagnf .
the figures show that over the relevant energy range , from the lyman limit to above the energy at which iron may be fully ionized ( 0.01 to @xmath13020 kev ) , oxaf predictions are sufficiently close to optxagnf predictions for our purposes ( differences are far less than modeling uncertainties ) .
the interquartile range displayed in each figure indicates that over the relevant energy range , oxaf predictions tend to be within @xmath13010 - 20% of the optxagnf value for the bbb component , and within @xmath131 for the non - thermal component .
the largest deviations tend to occur for energies which are probably too high to have a strong impact on photoionization model results ( 1 - 10 kev ) or are in an unlikely part of the parameter space ( e.g. @xmath132 @xmath18 )
. the most extreme discrepancies ( i.e. 95th percentile and upwards ) for the bbb comparisons tend to be due to irrelevant cases where only a small fraction of the total flux is in the relevant ionizing energy range , and hence the normalization over this range inflates a small section of the spectrum and magnifies proportional differences between the two models . for the non - thermal component
the differences are due only to the small scatter in the fit given in equation [ eq : t_r_cor ] , i.e. the seed blackbody spectra for oxaf and optxagnf have slightly different temperatures .
the oxaf spectra reproduce the optxagnf models with sufficient accuracy for photoionization modeling of optical diagnostic emission lines .
the differences between oxaf and optxagnf are smaller than the uncertainties due to , for example , any of the following : * the angular dependence of the accretion disk spectrum , including the effects of special relativistic beaming , gravitational beaming / light bending , and limb darkening @xcite .
gravitational redshifts are an additional consideration . * uncertainties in fundamental aspects of the thin disk model , such as the @xmath59-prescription , and aspects of the geometry of the accretion disk and corona including the variation with black hole spin ( e.g. * ? ? ?
the luminosity range over which the model is applicable is another consideration , e.g. for luminosities over @xmath6 @xmath133 a ` slim ' disk is a more appropriate model . * radiative transfer effects ; for example , the use of a color correction in the optxagnf model neglects h and he edges ( see figure 1 in * ? ? ?
* ) two simple tests were performed to compare mappings 5.1 photoionization models based on optxagnf and oxaf spectra .
for the first test the input optxagnf spectrum was configured with @xmath134 @xmath8 , @xmath118 , @xmath135 @xmath136 @xmath137 @xmath82 , @xmath138 @xmath18 , 2.2 ) , while the oxaf spectrum was configured with @xmath139 @xmath1 , @xmath140 ( 68.3 ev , 2.2 , 0.223 ) , where @xmath0 was calculated from equation [ eq_e_peak ] and @xmath2 was calculated to match the proportion in the optxagnf spectrum . the plane - parallel mappings models for this test were configured with a metallicity of @xmath141 , an ionization parameter of @xmath142 , and a constant pressure of @xmath143 k@xmath144@xmath145 .
the resulting line fluxes were broadly consistent between the two models .
of 43 optical lines with predicted fluxes of more than 1% of h@xmath146 , all had a flux discrepancy of less than 10% of the flux predicted using the optxagnf model .
all but 4 line fluxes showed a difference of less than 5% ; these 4 lines from or had differences of @xmath147 .
the reason for this discrepancy is that these lines from low - ionization species are very sensitive to small differences in the shape of the hard photon spectrum since the emission is produced mostly in a partially ionized zone heated by auger electrons .
for the second test the input optxagnf spectrum was configured with @xmath134 @xmath8 , @xmath118 , @xmath135 @xmath148 @xmath149 @xmath82 , @xmath150 @xmath18 , 1.8 ) , while the oxaf spectrum was configured with @xmath139 @xmath1 , @xmath140 ( 19.9 ev , 1.8 , 0.342 ) .
the mappings models used a metallicity of @xmath151 , an ionization parameter of @xmath152 , and a constant pressure of @xmath143 k@xmath144@xmath145 . with the somewhat softer ionizing spectrum and different nebular parameters ,
there were only 31 optical lines with predicted fluxes of more than 1% of h@xmath146 , and all of those lines had a flux discrepancy below @xmath153 , with 24 having a discrepancy under @xmath154 .
the accuracy of oxaf is more than sufficient for our purposes of photoionization modeling .
a grid of mappings photoionization models was used to explore the effect of the oxaf parameters on emission - line ratios on the standard optical diagnostic diagrams .
the grid was run over a range of photoionization model parameters and a range of all three oxaf parameters .
the results are shown in figure [ fig : mappings_grid ] for the dusty , plane - parallel mappings models configured with metallicity @xmath155 @xmath156 , ionization parameter @xmath157 , and a constant pressure of @xmath158 k@xmath144@xmath145 .
each model was iteratively re - run three times to ensure that the total pressure ( including radiation pressure ) was close to the input pressure , with this constraint applied where the model nebula was @xmath159 ionized .
the iterations for each gridpoint culminated in a model which ended when the nebula was @xmath160 neutral .
also plotted in figure [ fig : mappings_grid ] are gray points showing the measured spatially - resolved line flux ratios for the galaxy ngc 1365 from the s7 galaxy survey @xcite .
each point represents the line ratios for the narrowest gaussian kinematic component in the spectrum of a 1``@xmath1611 '' spatial pixel ( the seeing fwhm was @xmath162 " ) .
a s / n cut of 10 was applied to the linear flux ratios . in the leftmost
two columns of diagrams these points form a ` mixing sequence ' from line - ratios associated with excitation purely due to star formation ( at the bottom ) to excitation purely by agn radiation ( at the top ) .
the predicted line ratios for our fiducial mappings models approximately coincide with some of the measured ` pure - agn ' line ratios , especially in the leftmost column .
predictions of the weaker [ ] @xmath163 line however are inconsistent with the observations .
this line is produced predominantly in the partially ionized zone that arises due to the hard euv and x - ray photons .
the temperature and size of this region and thus the strength of the [ ] line are sensitive to the hard ionizing spectrum .
in addition the fractional ionization state of oxygen is strongly dependent upon charge exchange reactions and as a result this line is very difficult to predict accurately .
the [ ] @xmath164 6716 , 6731 doublet can also arise from this partially ionized region and is also somewhat inconsistent with the observations , albeit to a lesser degree .
the overpredicted fluxes from these species could to some extent be due to the nlr clouds being mass - bounded as opposed to ionization - bounded ( as in the models ) , i.e. more of the high - energy photons escape and the partially - ionized zones are shorter for the observed ensemble of nlr clouds in ngc 1365 compared to the models . in this section
we discuss figure [ fig : mappings_grid ] and interpret the effect of each oxaf parameter on the model predictions .
the logarithmically - spaced values of @xmath73 in figure [ fig : mappings_grid ] show that increasing the energy of the peak of the disk emission , and thereby increasing the energy of the ionizing photons , increases both the relative ionization state and the temperature of the nebula .
the sensitivity of optical diagnostic line ratios to @xmath73 is comparable to the sensitivity to the ionization parameter .
we first consider the expected range of values of @xmath107 .
@xcite study a sample of 51 unobscured seyfert 1 galaxies and fit the full three - component optxagnf model to the observed seds .
the proportion of the bolometric luminosity emitted in the power - law component in the best - fit model is less than 0.05 for @xmath166 of the sample , less than 0.2 for @xmath167 of the sample , and less than 0.4 for @xmath168 of the sample .
although the soft excess is a significant component ( allocated more flux than the power - law component for @xmath169 of the sample ) , it is effectively an extension of accretion disk emission to higher energies , so the proportion of flux in the power - law tail is comparable to the @xmath107 oxaf parameter .
hence we take @xmath170 as an appropriate range for typical ionizing agn spectra .
the values @xmath171 were included in the model grid but are not shown in figure [ fig : mappings_grid ] .
the diagnostic line - ratio predictions for @xmath172 were intermediate between those of @xmath173 and @xmath174 and were similar to those of @xmath173 .
the value @xmath175 was very high , with only a few extreme objects in the @xcite sample having such power - law - dominated seds .
hence the figure requires only the values @xmath176
@xmath177 to indicate the effect of @xmath107 and show the approximate magnitude of strong line - ratio variations due to @xmath107 variations that may occur between seyfert galaxies . increasing the value of @xmath107 increases the length of the trailing partially ionized zone in the photoionization models .
for @xmath73 = 32 ev , @xmath1 = 1.8 , @xmath178 u = -2.0 and z = 1.5 z@xmath179 , variation of @xmath107 from 0.1 to 0.7 increased the length of the modeled cloud by a factor of @xmath180 to 1.5 pc .
extending the partially - ionized zone increases the [ ] /h@xmath59 and [ ] /h@xmath59 line ratios , as the figure shows for @xmath181 . for values of @xmath182 , the predicted emission - line ratios do not show strong sensitivity to @xmath1 .
figure [ fig : mappings_grid ] shows how as @xmath1 increases , [ ] and [ ] line emission is enhanced relative to h@xmath59 . increasing @xmath1 allocates more of the non - thermal energy to relatively soft x - ray photons which are more easily absorbed by the nebula , causing more emission in the partially ionized zone .
hence @xmath1 has a similar effect to @xmath107 . as @xmath107 increases , the sensitivity to @xmath1 increases due to the greater relative effect of x - ray power - law photons on the nebula ionization structure .
in this section we compare oxaf spectra with the default spectrum produced by the agn spectral model supplied with the cloudy @xcite photoionization code .
we also compare the predicted optical diagnostic emission - line ratios produced by using these ionizing spectra in mappings photoionization models .
the cloudy agn spectral model uses an analytic formula consisting of exponential and power - law factors and terms .
the default cloudy agn model was configured with the following default values : a bbb temperature of @xmath183 k , an x - ray to uv ratio of @xmath184 , a low - energy bbb slope of @xmath185 , and an x - ray power - law index of @xmath186 ( corresponding to @xmath187 ) .
figure [ fig : cloudy_oxaf_spec ] shows the cloudy default sed , along with two ` equivalent ' oxaf spectra .
the first oxaf spectrum has @xmath139 @xmath1 , @xmath140 ( 6.9 ev , 2.0 , 0.46 ) , chosen to match the bbb peak energy and the ` height ' of the power - law tail , and has the same normalization as the normalized cloudy default sed .
the second ` equivalent ' oxaf spectrum has @xmath139 @xmath1 , @xmath140 ( 9.9 ev , 2.0 , 0.414 ) to match the shape and position of the high - energy slope of the bbb and match the ` height ' of the power - law tail .
this second oxaf spectrum is shown with a different normalization , with only 73% of the energy flux of the cloudy spectrum , to show how it follows the shape of the cloudy spectrum in the important region associated with h - ionizing fuv photons .
.,scaledwidth=48.0% ] clear differences between the spectral models are apparent in figure [ fig : cloudy_oxaf_spec ] .
the cloudy model uses a single blackbody curve combined with a power - law to produce the bbb , resulting in a bbb much wider than the oxaf bbb . the physical model used to calculate the power - law tail in oxaf produces a power - law cutoff that is much steeper than the simple piecewise @xmath188 cutoff in the cloudy model .
results of mappings photoionization model runs using the three spectra in figure [ fig : cloudy_oxaf_spec ] are presented in figure [ fig : cloudy_oxaf_bpt ] .
the plane - parallel dusty mappings models were all configured with a metallicity of @xmath189 , an ionization parameter of @xmath190 , and a constant pressure of @xmath143 .
all three models produced similar optical diagnostic line ratios .
the second oxaf model resulted in line ratios closer to those due to the cloudy default spectrum , which confirms that the most important part of the spectrum for ionization modeling is the region immediately on the high - energy side of the h ionization threshold , where these spectra were very similar by design .
the first oxaf model results in relatively lower [ ] @xmath191h@xmath146 , which is presumably a consequence of the lower proportion of hydrogen - ionizing euv photons .
the results show that oxaf spectra may produce predicted diagnostic line ratios that are very similar to those resulting from analytic ionizing spectra not based on physical models , provided the spectral shapes are sufficiently well - matched in the most important part of the spectrum , the hydrogen - ionizing euv .
we present a model of agn continuum emission , oxaf , designed for use in photoionization modeling and featuring spectral shapes based on physical models .
the oxaf model removes degeneracies between agn parameters in terms of their effect on the ionizing spectral shape ; in particular , in the thin disk accretion model the black hole mass and agn luminosity are entirely degenerate with respect to their impact on the spectral shape .
we base oxaf on the optxagnf model @xcite , and explain and remove some additional spectral - shape degeneracies between agn parameters when we reparametrize optxagnf .
we show that the soft x - ray excess is not important to photoionization modeling of the standard optical strong diagnostic lines , so we do not include it in the oxaf model
. however the soft excess must be important in modeling forbidden high - ionization lines .
the model oxaf contains only three parameters - the energy of the peak of the accretion disk emission @xmath0 , the photon power - law index for the non - thermal comptonized component @xmath1 , and the proportion of the total flux which goes to the non - thermal component @xmath107 .
these parameters intuitively describe the physical shape of the produced spectrum .
we show that predicted ratios of strong lines on standard optical diagnostic diagrams are sensitive to all three oxaf parameters .
the parameter @xmath0 directly affects the degree of ionization of the mappings model nebulae .
the parameters @xmath1 and @xmath107 are similar in their effects in that they change the length of the partially - ionized zone . predicted line - ratios are more sensitive to @xmath1 as @xmath107 is increased .
measured strong emission - line ratios for the seyfert galaxy ngc 1365 are approximately consistent with the predictions of some fiducial photoionization models using input oxaf ionizing spectra .
b.g . gratefully acknowledges the support of the australian research council as the recipient of a future fellowship ( ft140101202 ) .
l.k . and m.d .
acknowledge support from arc discovery project # dp160103631 .
c.j . acknowledges the support by the bundesministerium fr wirtschaft und technologie / deutsches zentrum fr luft- und raumfahrt ( bmwi / dlr , fkz 50 or 1604 ) and the max planck society .
we are grateful to both chris done , who received an early draft of this paper , and the anonymous referee for helpful comments .
allen , m. g. , dopita , m. a. , & tsvetanov , z. i. 1998 , , 493 , 571 antonucci , r. 1993 , , 31 , 473 baldwin , j. a. , phillips , m. m. , & terlevich , r. 1981 , , 93 , 5 bianchi , s. , guainazzi , m. , & chiaberge , m. 2006 , , 448 , 499 bland - hawthorn , j. , maloney , p. r. , sutherland , r. s. , & madsen , g. j. 2013 , , 778 , 58 collins , n. r. , kraemer , s. b. , crenshaw , d. m. , bruhweiler , f. c. , & melndez , m. 2009 , , 694 , 765 contini , m. , & viegas , s. m. 2001 , , 132 , 211 czerny , b. , you , b. , kurcz , a. , et al .
2016 , arxiv:1601.02498 davies , r. l. , dopita , m. a. , kewley , l. , et al .
2016 , , 824 , 50 davis , s. w. , done , c. , & blaes , o. m. 2006 , , 647 , 525 done , c. , davis , s. w. , jin , c. , blaes , o. , & ward , m. 2012 , , 420 , 1848 de marco , b. , ponti , g. , cappi , m. , et al .
2013 , , 431 , 2441 dopita , m. a. , groves , b. a. , sutherland , r. s. , binette , l. , & cecil , g. 2002 , , 572 , 753 dopita , m. a. , scharwchter , j. , shastri , p. , et al .
2014 , , 566 , a41 dopita , m. a. , shastri , p. , davies , r. l. , et al .
2015 , , 217 , 12 evans , i. , koratkar , a. , allen , m. , dopita , m. , & tsvetanov , z. 1999 , , 521 , 531 fabian , a. c. 2016 , astronomische nachrichten , 337 , 375 ferland , g. j. , porter , r. l. , van hoof , p. a. m. , et al . 2013 , rmxaa , 49 , 137 gierliski , m. , & done , c. 2004 , , 349 , l7 groves , b. a. , dopita , m. a. , & sutherland , r. s. 2004 , , 153 , 9 groves , b. a. , heckman , t. m. , & kauffmann , g. 2006 , , 371 , 1559 jin , c. , ward , m. , done , c. , & gelbord , j. 2012 , , 420 , 1825 jin , c. , ward , m. , & done , c. 2012 , , 422 , 3268 jin , c. , ward , m. , & done , c. 2012 , , 425 , 907 kauffmann , g. , heckman , t. m. , tremonti , c. , et al .
2003 , , 346 , 1055 kewley , l. j. , dopita , m. a. , sutherland , r. s. , heisler , c. a. , & trevena , j. 2001 , , 556 , 121 kewley , l. j. , groves , b. , kauffmann , g. , & heckman , t. 2006 , , 372 , 961 laor , a. , & netzer , h. 1989 , , 238 , 897 lightman , a. p. , & zdziarski , a. a. 1987 , , 319 , 643 molina , m. , bassani , l. , malizia , a. , et al .
2013 , , 433 , 1687 murayama , t. , & taniguchi , y. 1998 , , 503 , l115 novikov , i. d. , & thorne , k. s. 1973 , black holes ( les astres occlus ) , 343 page , d. n. , & thorne , k. s. 1974 , , 191 , 499 pal , m. , dewangan , g. c. , misra , r. , & pawar , p. k. 2016 , , 457 , 875 reynolds , c. s. 2014 , , 183 , 277 shakura , n. i. , & sunyaev , r. a. 1973 , , 24 , 337 veilleux , s. , & osterbrock , d. e. 1987 , , 63 , 295 viegas - aldrovandi , s. m. , & contini , m. 1989 , , 215 , 253 vika , m. , driver , s. p. , graham , a. w. , & liske , j. 2009 , , 400 , 1451 walton , d. j. , nardini , e. , fabian , a. c. , gallo , l. c. , & reis , r. c. 2013 , , 428 , 2901 you , b. , cao , x. , & yuan , y .- f .
2012 , , 761 , 109 zdziarski , a. a. , johnson , w. n. , & magdziarz , p. 1996
, , 283 , 193 ycki , p. t. , done , c. , & smith , d. a. 1999 , , 309 , 561 kev .
the statistics in each spectral bin were taken over oxaf - optxagnf comparisons at @xmath192 points in the parameter space , and are plotted as a function of bin energy ( left column ) , and bin energy normalized by the energy of the peak of the disk emission @xmath0 ( right column ) . for subplots in the top row , a bin in a spectrum was only included in the statistics if the optxagnf bin flux was more than 0.1% of the maximum optxagnf flux for a bin in the range @xmath193 kev .
the accuracy of oxaf in reproducing the optxagnf spectra is more than adequate over the relevant energy range of @xmath194 kev .
, scaledwidth=100.0% ] kev .
the statistics in each spectral bin were taken over oxaf - optxagnf comparisons at @xmath192 points in the parameter space , and are plotted as a function of bin energy ( left column ) , and bin energy normalized by the energy of the peak of the disk emission @xmath0 ( right column ) . for subplots in the top row , a bin in a spectrum was only included in the statistics if the optxagnf bin flux was more than 0.1% of the maximum optxagnf flux for a bin in the range @xmath193 kev .
the accuracy of oxaf in reproducing the optxagnf spectra is more than adequate over the relevant energy range of @xmath194 kev.,scaledwidth=100.0% ] | we present a simplified model of active galactic nucleus ( agn ) continuum emission designed for photoionization modeling .
the new model oxaf reproduces the diversity of spectral shapes that arise in physically - based models .
we identify and explain degeneracies in the effects of agn parameters on model spectral shapes , with a focus on the complete degeneracy between the black hole mass and agn luminosity .
our re - parametrized model oxaf removes these degeneracies and accepts three parameters which directly describe the output spectral shape : the energy of the peak of the accretion disk emission @xmath0 , the photon power - law index of the non - thermal emission @xmath1 , and the proportion of the total flux which is emitted in the non - thermal component @xmath2 .
the parameter @xmath0 is presented as a function of the black hole mass , agn luminosity , and ` coronal radius ' of the optxagnf model upon which oxaf is based .
we show that the soft x - ray excess does not significantly affect photoionization modeling predictions of strong emission lines in seyfert narrow - line regions . despite its simplicity ,
oxaf accounts for opacity effects where the accretion disk is ionized because it inherits the ` color correction ' of optxagnf .
we use a grid of mappings photoionization models with oxaf ionizing spectra to demonstrate how predicted emission - line ratios on standard optical diagnostic diagrams are sensitive to each of the three oxaf parameters .
the oxaf code is publicly available in the astrophysics source code library . |
nowadays it is still an open question which is the true nature of neutron stars . in a traditional picture
the core of a neutron star is modeled as an uniform fluid of neutron - rich matter in equilibrium with respect to the weak interaction .
nevertheless , due to the large stellar central density new degrees of freedom such as hyperons , meson condensates or even a deconfined phase of quark matter are expected to appear in the inner core of the star @xcite .
the latter possibility is a consequence of the qcd , that predicts a phase transition from hadron matter to a deconfined quark phase at few times nuclear saturation density , and was realized by several researchers soon after the introduction of quarks as the fundamentals building blocks of hadrons @xcite .
berezhiani _ et al .
_ @xcite have shown recently that when finite effects at the interface between the quark and the hadron phase are taken into account , pure hs ( _ i.e. , _ without a phase of deconfined quarks ) , above a threshold value of the central pressure , are metastable to `` decay '' into a more compact star in which deconfined quark matter is present ( qs ) .
the mean - life time of the metastable star configurations is related to the time needed to form a drop of quark matter in the stellar center , and depends dramatically on the stellar central pressure . in this work
, we study the role played by neutrino trapping on this conversion mechanism .
the quantum nucleation of a quark matter drop inside hadron matter is briefly revised in sec .
[ sec : sec2 ] .
our main results are presented in sec .
[ sec : sec3 ] , whereas the main conclusions are given in sec .
[ sec : sec4 ] .
let us consider a pure hs whose central density ( pressure ) is increasing due to spin - down or due to mass accretion ( from a companion star or from the interstellar medium ) .
as the central density approaches the quark deconfinement critical density , a drop of quark matter ( qm ) can be formed in the central region of the star .
the process of formation of the drop is regulated by its quantum fluctuations in the potential well created from the difference between the energy densities of the hadron and quark phases .
keeping only the volume and the surface terms , the potential well takes the simple form @xmath0 where @xmath1 is the quark baryon number density , @xmath2 and @xmath3 are the quark and hadron chemical potentials at a fixed pressure @xmath4 , and @xmath5 is the surface tension for the surface separating the hadron from the quark phase . the time needed to form the drop ( nucleation time )
can be straightforwardly evaluated within a semiclassical approach @xcite .
first one computes , in the wentzel
kramers brillouin ( wkb ) approximation , the ground state energy @xmath6 and the oscillation frequency @xmath7 of the drop in the potential well @xmath8 .
then , the probability of tunneling is given by @xmath9 \label{eq : prob}\ ] ] where @xmath10 is the action under the potential barrier which in a relativistic framework reads @xmath11[u(r)-e ] } \ .
\label{eq : action}\ ] ] being @xmath12 the classical turning points and @xmath13 the droplet effective mass , with @xmath14 and @xmath15 the hadron energy density and the hadron baryon number density , respectively .
the nucleation time is then equal to @xmath16 where @xmath17 is the number of virtual centers of droplet formation in the star .
a simple estimation gives @xmath18 @xcite .
to describe both the hadron and the quark phases we have adopted rather common models .
the hadron phase is described within the relativistic mean field approximation with standard parameters fitted to reproduce nuclear matter saturation properties @xcite , whereas for the quark phase we have adopted a phenomenological equation of state ( eos ) based on the mit bag model for hadrons @xcite with three parameters : the mass of the strange quark @xmath19 , the so - called pressure of the vacuum @xmath20 ( bag constant ) and the qcd structure constant @xmath21 . in the present work ,
we take @xmath22 mev and @xmath23 ( see ref .
@xcite for details ) .
to begin with we show in fig . [
fig : figure1 ] the potential energy @xmath8 for the formation of a qm drop for different values of the stellar central pressure above the so - called static transition point @xmath24 ( the pressure at which the hadron and quark chemical potentials are equal ) .
the surface tension @xmath5 is taken equal to @xmath25 mev/@xmath26 and @xmath27 mev/@xmath28 .
right ( left ) panel corresponds to the case in which the presence of neutrinos has ( not ) been taken into account in the eos of the hadron and quark phases . as expected the potential barrier is lowered as the central pressure increases .
note that the static transition point @xmath24 is larger when neutrinos are present , and note also that in this case , higher values of the central pressure are necessary to get a potential barrier with a height similar to the one obtained when neutrinos are absent .
this means that for a given value of the central pressure the nucleation time will be much larger when neutrinos are present , and therefore , the conversion of a hs to a qs will be inhibit in this case ( see figs . [
fig : figure2 ] and [ fig : figure3 ] ) . the nucleation time can be plotted as a function of the baryonic mass of the hs corresponding to the given value of the central pressure , as implied by the solution of the tolman - oppenheimer - volkov equations for the pure hs sequences . the results of our calculation for the two cases ( neutrino free and trapped ) are reported in fig . [
fig : figure2 ] for the same values of @xmath20 and @xmath5 of fig .
[ fig : figure1 ] .
as it can be seen from the figure a metastable hs can have a mean - life time ( related to the nucleation time ) many orders of magnitude larger than the age of the universe .
nevertheless , as the star accreates a small amount of mass ( @xmath29 few percent of @xmath30 ) , the consequential increase of the central pressure leads to a huge reduction of the nucleation time and , as a result , to a dramatic reduction of the hs mean - life time . from the discussion of the previous figure , it is clear that if neutrinos are trapped the star will need to have a much larger baryonic mass ( or central pressure ) than the star without neutrinos to have a comparable mean - life time . in the particular case of a star with a mean - life time @xmath31 year
the baryonic mass of the star with neutrinos is @xmath32 , whereas if the neutrinos have diffused out of the star @xmath33 . finally , in fig .
[ fig : figure3 ] we show the mass - radius ( mr ) relation for the initial hs and final qs sequences for the neutrino free ( left panel ) and neutrino trapped ( right panel ) cases . in both panels
the configuration marked with an asterisk on the hadron mr curves represents the hs for which the central pressure is @xmath24 .
all the configurations below this point correspond to absolutely stable hs ( @xmath34 ) .
the full circles ( triangle ) on the hs and qs sequences of the neutrino free ( trapped ) case represent the hs for which @xmath31 year and the final qs which is formed from its conversion ( in the neutrino trapped case deconfinement leads to the formation of a black hole for this particular set of parameters ) .
the square in the neutrino trapped case corresponds to the hs configuration which will evolve ( assuming baryon number conservation ) to the hs with @xmath31 year in the neutrino free case as the neutrinos diffuse out of the star .
it is clear from the figure that this and almost all the hs configurations in the neutrino trapped case will be stable with respect to the conversion .
in the present work we have studied the role of neutrino trapping in the hs to qs conversion mechanism proposed in ref . @xcite .
we have found that the quantum nucleation of a quark matter drop inside hadron matter , and therefore the conversion of a hs into a qs , is strongly inhibit by the presence of neutrinos .
d. ivanenko and d. f. kurdgelaidze , nuovo cimento 2 ( 1969 ) 13 ; n. itoh , prog .
theor . phys .
44 ( 1970 ) 291 ; f. iachello , w. d. langer and a. lande , nucl .
phys . a 219 ( 1974 ) 612 : j. c. collins and m. j. perry , phys .
34 ( 1975 ) 1353 . | we study the role played by neutrino trapping on the hadron star ( hs ) to quark star ( qs ) conversion mechanism proposed recently by berezhiani and collaborators .
we find that the nucleation of quark matter drops inside hadron matter , and therefore the conversion of a hs into a qs , is strongly inhibit by the presence of neutrinos . |
describing a fluid flow , one labels particles of the fluid by means of lagrangian coordinates @xmath0 .
one supposes , that any lagrangian coordinates @xmath1 label the same fluid particle all the time . applying laws of newtonian dynamics to any fluid particle , one obtains hydrodynamic equations for the fluid flow in the lagrangian representation ( in the lagrangian coordinates ) .
the lagrangian representation is sensitive to the correct labeling of the fluid particles in the sense , that there are situations , when in different time moments the same lagrangian coordinates @xmath1 describe different fluid particles . for instance , if two like gas beams , consisting of noninteracting molecules , pass one through another , the particle labeling changes after `` collision '' of the two beams .
the picture is shown in the figure , which describes world lines of gas particles in the space - time .
the solid lines shows gas particle with the same labeling @xmath1 , whereas dashed lines show world lines of real gas molecules .
in this example a violation of the fluid particle labeling after `` collision '' is evident .
the stream lines , represented by solid lines , do not describe motion of real fluid particles .
is it important ?
can we observe stream lines of a fluid ?
let us imagine that we introduce several flecks of dust in one of gas flows and follow their motion .
we suppose that the size of flecks is larger , than the size of gas molecules , and any fleck interacts with many gas molecules .
then we may think , that any fleck of dust moves along stream line of the fluid .
as far as collision of any fleck of dust with gas molecules is random , the flecks of dust after `` collision '' appear in both gas beams , although before the `` collision '' they were placed only in one of them . on the other hand ,
it is generally assumed , that flecks of dust moves together with the gas , and any fleck moves along the stream line of the fluid .
observation of flecks of powder is a usual method of the stream lines investigation .
the hydrodynamic description of a fluid is valid at the supposition , that one stream of a fluid can penetrate into the other one only to the depth of the mean length of collision path . in this case
the interfusion of different streams of a fluid will be infinitesimal .
however , such a small interfusion will take place , and this interfusion may appear to be essential for the shape of stream lines .
such a physical phenomenon as turbulence can be discovered only , if one traces the irregular behavior of stream lines . in other words , for observation of turbulence a displacement of fluid particles is important , but not only their velocities .
the velocities are important only as a source of displacement .
a motion of the inviscid barotropic fluid is described by the euler equations @xmath2 where @xmath3 is the fluid density , @xmath4 is the fluid velocity , @xmath5 is the pressure and @xmath6 is the fluid internal energy per unit mass .
stream lines are described by a system of ordinary differential equations .
one supposes that fluid particles move along the stream lines @xmath7 , which are defined by the equation @xmath8 where @xmath4 is a solution of the euler equations ( [ a1.1 ] ) , ( [ a1.2 ] ) . the euler system ( [ a1.1 ] ) , ( [ a1.2 ] ) is a closed system of differential equations , which may be solved independently of equations ( a1.3 ) .
the euler system is a system of nonlinear partial differential equations .
it is difficult for solution .
the system of ordinary differential equations ( [ a1.3 ] ) is simpler , than the euler system .
besides , it can be solved only after solution of the euler system ( [ a1.1 ] ) , ( [ a1.2 ] ) .
it is a reason , why researchers investigate mainly the euler system .
the system of equations ( [ a1.3 ] ) for the stream lines remains usually outside the region of consideration as some triviality .
however , the euler system ( [ a1.1 ] ) , ( [ a1.2 ] ) and equations ( a1.3 ) are dynamic equations of one dynamic system , and they should be considered together . this dynamic system will be referred as the complete hydrodynamic system .
the euler equations ( [ a1.1 ] ) , ( [ a1.2 ] ) are not dynamic equations of a wholesome dynamic system , because the can not be deduced from a variational principle , whereas dynamic equations ( [ a1.1 ] ) , ( [ a1.2 ] ) , ( [ a1.3 ] ) can .
it seems , that dynamic equations do not influence on the solution of dynamic equations ( [ a1.1 ] ) , ( [ a1.2 ] ) . in reality , it is true only for irrotational flows . in vortical
flows a situation changes . in
the vortical flows an interfusion appears .
the interfusion is not so large as in figure with colliding gas beams .
this interfusion is infinitesimal .
it is conditioned by different velocities of adjacent fluid volumes .
this infinitesimal interfusion influences on the shape of stream lines and on the labeling of the fluid particles , although it does not influence on quantities @xmath9 , which are solution of the euler system ( [ a1.1 ] ) , ( [ a1.2 ] ) . in the case of the euler representation of hydrodynamic equations ,
the fluid particle labeling is not necessary .
the four hydrodynamic euler equations are obtained as a result of the conservation laws of the energy and momentum . however , in this case the conservation law of the angular momentum is not used , and one can not be sure , that the system of four euler equations for barotropic fluid describes completely rotational degrees of freedom of molecules and those of fluid particles .
the rotational degrees of freedom may be essential in turbulent flows , where close consideration of rotational degrees of freedom may appear to be essential .
in this paper we try to take into account interfusion , which appears in the rotational flows of the barotropic fluid .
influence of interfusion manifests itself in labeling of the fluid particles by means of the lagrangian coordinates @xmath10 .
this change of labeling is not important for solution of euler equations ( [ a1.1 ] ) , ( [ a1.2 ] ) , which form a closed system of differential equations .
however , this change of labeling may appear to be important for such physical phenomena , where the shape of stream lines is essential ( such as turbulence ) .
we consider connection between the labeling and the interfusion on the formal mathematical level .
we shall consider the euler dynamic equation for inviscid barotropic fluid we are interesting in the question , whether the labeling , generated by the equation ( [ a1.3 ] ) is an unique possible way of labeling . to solve this problem
, we shall consider the lagrangian coordinates @xmath1 to be dependent dynamic variables .
the eulerian coordinates @xmath11 are considered to be independent dynamic variables .
thus , we consider dynamic system , described by seven dependent dynamic variables @xmath12 * * , * * which are functions of four independent variables @xmath11 .
note that system of dynamic equations ( [ a1.1 ] ) , ( [ a1.2 ] ) is a closed system of dynamic equations .
however , the dynamic system , described , by four dependent variables @xmath9 is not a wholesome dynamic system in the sense , that dynamic equations ( a1.1 ) , ( [ a1.2 ] ) can not be obtained from some a variational principle . to obtain dynamic equations ( [ a1.1 ] ) , ( [ a1.2 ] ) from the variational principle @xcite , one needs to add so - called lin constraints @xcite .
this conditions have the form @xmath13it is easy to see , that characteristics of the linear differential equation ( [ b1.3])@xmath14coincide with the equation ( [ a1.3 ] ) .
vice versa , any integral of the equation system ( [ a1.3 ] ) is a solution of the equation ( [ b1.3 ] )
. the lin constraints ( [ b1.3 ] ) are interesting in the relation , that independent dynamic variables in ( [ b1.3 ] ) are the same , as in the dynamic equations ( [ a1.1 ] ) , ( [ a1.2 ] ) .
hence , dynamic equations ( a1.1 ) , ( [ a1.2 ] ) and dynamic equations ( [ b1.3 ] ) may be considered as dynamic equations of one dynamic system .
it is rather difficult to consider system of equations ( [ a1.1 ] ) , ( [ a1.2 ] ) , ( [ a1.3 ] ) as a dynamic equations of a dynamic system , because independent variables are different in equations ( [ a1.1 ] ) , ( [ a1.2 ] ) and ( [ a1.3 ] ) .
let us note that the quantities @xmath1 may be considered to be the generalized stream function ( gsf ) , because @xmath1 have two main properties of the stream function .
gsf @xmath1 labels stream lines of a fluid .
some combinations of the first derivatives of any @xmath1 satisfy the continuity equation identically.@xmath15where @xmath16 is the 4-vector of flux .
here and in what follows , a summation over two repeated indices is produced ( 0 - 3 ) for latin indices and ( 1 - 3 ) for greek ones .
the jacobian determinant @xmath17 @xmath18is considered to be a four - linear function of @xmath19 .
the quantity @xmath20 is the temporal lagrangian coordinate , which appears to be fictitious in expressions for the flux 4-vector @xmath21 @xmath22 a use of jacobians in the description of the ideal fluid goes up to clebsch ( * ? ? ?
* ; * ? ? ?
* clebsch , 1857,1859 ) , who used jacobians in the expanded form .
it was rather bulky .
we use a more rational designations , when the 4-flux and other essential dynamic quantities are presented in the form of derivatives of the principal jacobian @xmath23 . dealing with the generalized stream function @xmath24 ,
the following identities are useful @xmath25@xmath26see details of working with jacobians and the generalized stream functions in ( * ? ? ?
* rylov,2004 ) . _
example_. _
application of the stream function for integration of equations , describing the 2d stationary flow of incompressible fluid . _ dynamic equations have the form@xmath27where @xmath28 and @xmath29 are velocity components along @xmath30-axis and @xmath31-axis respectively . introducing the stream function @xmath32 by means of relations@xmath33we
satisfy the first equation ( [ a2.6 ] ) identically , and we obtain for the second equation ( [ a2.6 ] ) the relations@xmath34which can be rewritten in the form@xmath35where @xmath36 is the vorticity of the fluid flow .
the general solution of equation ( [ a2.8 ] ) has the form@xmath37where @xmath38 is an arbitrary function of @xmath32 . for the irrotational flow the vorticity @xmath39 ,
and we obtain instead ( [ a2.9 ] ) @xmath40 one obtains the unique solution of ( [ a2.10 ] ) inside of a closed region of 2d space provided , that the value @xmath41 of the stream function @xmath32 is given on the boundary @xmath42 of this region .
the differential structure of equations ( [ a2.9 ] ) and ( [ a2.10 ] ) is similar .
one should expect , that giving the value @xmath41 of the stream function @xmath32 on the boundary @xmath42 , one obtains the unique solution of the equation ( [ a2.10 ] ) .
but it is not so , because the indefinite function @xmath43 is not given , and it can not be determined from the boundary condition , because the nature of the function @xmath43 is another , than the nature of the boundary conditions .
first , if the flow contains closed stream lines , which do not cross the boundary , one can not determine the values of @xmath38 on these stream lines from the boundary conditions .
but for determination of the unique solution the values of @xmath38 on the closed stream lines must be given .
second , the boundary conditions are given arbitrarily .
the function @xmath38 can not be given arbitrarily .
for those stream lines , which cross the boundary more than once , the values of @xmath38 on the different segments of the boundary are to be agreed .
thus , the nonuniqueness of the solution , connected with the indefinite function @xmath38 has another nature , than the nonuniqueness , connected with the insufficiency of the boundary conditions .
we use the variational principle for the derivation of the hydrodynamic equations ( [ a1.1 ] ) , ( [ a1.2 ] ) , ( [ b1.3 ] ) .
the action functional has the form@xmath44 = \dint\limits_{v_{x}}\left\ { \frac{\mathbf{j}^{2}}{2\rho } -\rho e\left ( \rho \right ) -p_{k}\left ( j^{k}-\rho _ { 0}\left ( \mathbf{\xi } \right ) \frac{\partial j}{\partial \xi _ { 0,k}}\right ) \right\ } d^{4}x , \label{a3.1}\]]where @xmath45 , @xmath46 are the lagrange multipliers , which introduce the designations for the 4-flux@xmath47note , the expression for the 4-flux ( [ a3.2 ] ) satisfies the first equation ( [ a2.1 ] ) identically , because the expression ( [ a3.2 ] ) may be reduced to the form of the second relation ( [ a2.1 ] ) by means of a change of variables @xmath1@xmath48then@xmath49 besides according to the first identity ( [ a2.4 ] ) the relation ( [ a3.2 ] ) satisfies the lin constraint ( [ b1.3 ] ) .
variation of the action ( [ a3.1 ] ) with respect to @xmath50 gives relations ( [ a3.2 ] ) .
another dynamic equations have the form@xmath51@xmath52@xmath53using the third relation ( [ a2.4 ] ) , we obtain@xmath54 now using ( [ a2.5 ] ) , we obtain @xmath55using the first relation ( [ a2.4 ] ) , we obtain@xmath56 there are two ways of dealing with this equation : \1 . elimination of gsf @xmath1 , which leads to the euler equations .
integration , which leads to appearance of arbitrary functions .
_ the first way : elimination of gsf _ convoluting ( [ a3.8 ] ) with @xmath57 and using dynamic equations ( a3.2 ) , we obtain @xmath58 substituting @xmath45 and @xmath21 from relations ( [ a3.3 ] ) and ( [ a3.4 ] ) , we obtain the euler dynamic equations ( [ a1.1])@xmath59the continuity equation ( [ a1.2 ] ) is a corollary of equations ( [ a3.2 ] ) and identity ( [ a2.1 ] ) . finally the lin constraints ( [ b1.3 ] ) are corollaries of the first identity ( [ a2.4 ] ) and dynamic equations ( a3.2 ) . _ the second way : integration of the equation for _
@xmath45 let us consider the equations ( [ a3.8 ] ) as linear differential equations for @xmath45 .
the general solution of ( [ a3.8 ] ) has the form@xmath60where @xmath61 @xmath62 are arbitrary functions of @xmath1 , @xmath63 is a new variable instead of fictitious variable @xmath20 .
let us differentiate ( [ a3.11 ] ) and substitute the obtained expressions @xmath64 in ( [ a3.8 ] ) . using the first identity ( [ a2.4 ] ) , we see , that the relations ( [ a3.12 ] ) satisfy the equations ( [ a3.8 ] ) identically . we may substitute ( [ a3.11 ] ) in the action ( [ a3.1 ] ) , or introduce ( [ a3.11 ] ) by means of the lagrange multipliers .
( the result is the same ) .
we obtain the new action functional @xmath65 = \dint\limits_{v_{x}}\left\ { \frac{\mathbf{j}^{2}}{2\rho } -\rho e\left ( \rho \right ) -j^{k}\left ( \partial _ { k}\varphi + g^{\alpha } \left ( \mathbf{\xi } \right ) \partial _ { k}\xi _ { \alpha } \right ) \right\ } d^{4}x , \label{a3.14}\]]which contains arbitrary integration functions @xmath66 .
here@xmath67the integration functions @xmath68 are considered as a fixed functions of @xmath1 .
the term @xmath69is omitted , because it does not contribute to dynamic equations .
variation of ( [ a3.14 ] ) with respect to @xmath70 , @xmath71 and @xmath72 gives respectively@xmath73@xmath74@xmath75 variation of ( [ a3.14 ] ) with respect to @xmath76 gives@xmath77where@xmath78and @xmath79 is determined by the relation ( [ a3.19 ] ) if @xmath80 , then the lin constraints @xmath81follows from ( [ a3.20 ] ) however , the matrix @xmath82 is antisymmetric and @xmath83then it follows from ( [ a3.20])@xmath84where @xmath85 is an arbitrary quantity , and @xmath86 is the weight function from ( [ a3.2 ] ) . the obtained equation ( [ a3.24 ] ) contains the initial dynamic equation ( [ b1.3 ] ) as a special case . for irrotational flow , when @xmath87 , the equation ( [ a3.24 ] ) turns to ( [ b1.3 ] ) . in the action functional ( [ a3.1 ] ) the initial relation ( [ b1.3 ] ) is used as a side constraint .
it is a reason , why the equation ( [ a3.24 ] ) is not obtained from the action functional ( [ a3.1 ] ) .
note , that eliminating the variables @xmath70 and @xmath1 from dynamic equations ( [ a3.18 ] ) - ( [ a3.20 ] ) , we obtain the euler dynamic equations ( [ a1.1 ] ) .
the vorticity @xmath88 and @xmath89 are obtained from ( [ a3.19 ] ) in the form @xmath90@xmath91 let us form a difference between the time derivative of ( [ a3.19 ] ) and the gradient of ( [ a3.18 ] ) . eliminating @xmath92 from the obtained equation by means of equations ( [ a3.20 ] ) , one obtains @xmath93using ( [ b3.29 ] ) and ( [ b3.28 ] ) , the expression ( [ b3.30 ] ) reduces to @xmath94 in virtue of the identity @xmath95the equation ( [ b3.31 ] ) is equivalent to ( [ a1.1 ] ) .
note , that the euler equations ( [ a1.1 ] ) are obtained at any form of the arbitrary function @xmath96 in the equations ( [ a3.24 ] ) , because the equations ( [ a3.24 ] ) are used in the form ( a3.20 ) , where the form of @xmath96 is unessential .
solution of the euler system ( [ a1.1 ] ) , ( [ a1.2 ] ) in the form @xmath3 , @xmath97 does not depend on the form of the indefinite function @xmath96 . if @xmath98 , the dynamic equations ( a3.24 ) describe a violation of the lin constraints ( [ b1.3 ] ) .
one obtains another labeling of the stream lines , than that one , which is described by the lin constraints ( [ b1.3 ] ) .
if the flow is irrotational , and @xmath99 , the labeling does not depend on @xmath96 .
let us consider two different labeling @xmath1 and @xmath100 of the the same fluid flow described by the variables @xmath101 , @xmath4 .
the initial conditions are supposed to have the form@xmath102then according to ( [ a3.19 ] ) , ( [ a3.21])@xmath103according to ( [ a3.24 ] ) the dynamic equations for labeling @xmath1 and @xmath100 have the form @xmath104@xmath105where the velocity @xmath106 is the same in both equations and the function @xmath107 is defined by the relation @xmath108if the velocity @xmath79 is defined by relations ( [ a3.19 ] ) , it satisfies the euler equations and associates with the generalized stream function @xmath109 , whose evolution is described by the equations ( [ a3.24 ] ) in general , the evolution of the quantities @xmath100 and @xmath1 is different , although the coincide at @xmath110 .
let @xmath111it follows from ( [ b3.40 ] ) and ( [ b3.41 ] ) that mismatch @xmath112 between @xmath100 and @xmath1 is determined by the relation@xmath113the system of ordinary differential equations , associated with the equation ( [ a3.33 ] ) , has the form@xmath114 solution of the system of ordinary equations at the initial conditions @xmath115 has the form@xmath116then@xmath117 thus , although solution @xmath9 of the cauchy problem for the euler system of hydrodynamic equation ( [ a1.1 ] ) , ( [ a1.2 ] ) is unique , the solution @xmath12 for the cauchy problem of the complete system of hydrodynamic equations ( [ a1.1 ] ) , ( [ a1.2 ] ) , ( [ a3.24 ] ) is not unique .
the reason of this nonuniqueness is consideration of interfusion .
this consideration is formal .
one can not understand mechanism of the interfusion influence from this consideration .
nevertheless this influence takes place , and it should be investigated more closely .
it seems , that in the two - dimensional flow instead of determinant ( a3.23 ) we have the determinant@xmath118which does not vanish , in general .
then the problem of nonuniqueness of thelabeling is removed and the solution of the cauchy problem for the complete hydrodynamic system becomes to be unique . in reality , we may control the solution only via initial conditions .
we may give the two - dimensional initial conditions , i.e. @xmath119 in this case@xmath120the determinant@xmath121and the relations ( [ a3.24 ] ) take the form@xmath122one can not control indefinite quantity @xmath123 , which may depend on @xmath124 .
the equation ( [ a4.6 ] ) generates nonunique solution of the cauchy problem of vortical flow for the complete hydrodynamic system .
the flow with the two - dimensional initial conditions turns into three - dimensional vortical flow .
solution of the cauchy problem for the vortical flow of inviscid barotropic fluid is not unique , if we solve seven dynamic equations of the complete hydrodynamic system , which includes description of the shape of stream lines and has seven dependent variables @xmath71 , @xmath79 , @xmath1 .
nonuniqueness is connected with the fact , the initial conditions for variables @xmath71 , @xmath79 , @xmath1 do not control the intermixing effect .
solution of the cauchy problem for the vortical flow of inviscid barotropic fluid is unique , if we solve only four dynamic equations of the euler system and ignore shape of stream lines . in this case
dynamic equations are written for four dependent variables @xmath71 , @xmath79 . the intermixing effect , generating nonunique solutions , associates with the turbulence phenomenon at the following points : ( 1 ) both effects are not controlled by the initial data for variables @xmath71 , @xmath79 , @xmath1 of the hydrodynamic equations in the euler representation , ( 2 ) both effects take place at the vortical flows , and they are absent at the irrotational flows , ( 3 ) both effects are strong at vanishing viscosity .
we admit , that the interfusion may be connected with the turbulence phenomena , although we do not yet insist on this statement .
if , indeed , the turbulent phenomena are connected with the interfusion and with the shape of stream lines , it becomes clear , why numerous investigations of hydrodynamic equations , describing only density and velocity , but not the shape of stream lines , had not led to a progress .
the researchers looked for turbulence in that region , where it is not placed .
a. rylov , hydrodynamic equations for incompressible inviscid fluid in terms of generalized stream function .
_ int . j. math . & mat .
2004 * , no .
11 , 21 february 2004 , pp .
541 - 570 .
( available at http://arxiv.org/abs/physics/0303065 ) . | it is shown that the euler system of hydrodynamic equations for inviscid barotropic fluid for density and velocity is not a complete system of dynamic equations for the inviscicd barotropic fluid .
it is only a closed subsystem of four dynamic equation .
the complete system of dynamic equation consists of seven dynamic equations for seven dependent variables : density , velocity and labeling ( lagrangian coordinates , considered as dependent variables ) .
solution of the cauchy problem for the euler subsystem is unique .
solution of the cauchy problem for the complete hydrodynamic system , containing seven equations , is unique only for irrotational flows . for vortical flows solution of the cauchy problem
is not unique .
the reason of the nonuniqueness is an interfusion , which can not be taken into account properly in the framework of hydrodynamics .
there are some arguments in favour of connection between interfusion and turbulence . |
an application of quantum phenomena to securing optical network has received much attention . in this case , instead of mathematical encryption , a guarantee of security by a physical principle is expected .
so far the quantum key distribution(qkd ) based on very weak optical signals has been developed and demonstrated in many institutions .
however , these have inherent difficulty such as quantum implementations and very low bit rates compared to current data transmission rates . in order to cope with such a drawback , in 2000 ,
a new quantum cipher was proposed [ 1 ] .
it is a kind of stream cipher with randomization by quantum noise generated in measurement of coherent state from the conventional laser diode .
the scheme is called yuen-2000 protocol(y-00 ) or @xmath0 scheme[2,3 ] which consists of large number of basis to transmit the information bit and pseudo random number generator(prng ) for the selection of the basis . a coherent state as the ciphertext which is transmitted from the optical transmitter(alice ) is determined by the input data and the running key @xmath1 from the output sequence of prng with a secret key @xmath2 .
the legitimate receiver(bob ) has the same prng , so he can receive the correct ciphertext under the small error , and simultaneously demodulate the information bit .
the attacker ( eve ) , who does not know the key , has to try to discriminate all possible coherent state signals . since the signal distance among coherent state signals are designed as very small , eve s receiver suffers serious error to get the ciphertext .
such a difference of the accuracy of the ciphertext for bob and eve brings preferable security which can not be obtained in any conventional cipher .
unfortunately , it is very difficult to clarify the quantitative security evaluation for this type of cipher , because all physical parameter for the cipher system is finite .
so far , complexity theory approach [ 4 ] and information theoretic approach [ 5 ] have been tried , but still there is no rigorous theoretical treatment .
recently , yuen has pointed out that it is possible to show the rigorous security analysis when the parameters are allowed to be asymptotical , and showed a sketch of the properties using a model of coherent pulse position modulation ( cppm ) [ 6 ] .
this may open a new way for the quantum key distribution with coherent states of considerable energy and high speed . in this paper , we clarify a generation method of cppm quantum signal by using a theory of unitary operator and symplectic transformation , and show a security property and its numerical examples . in the section
ii , the back ground for the information theoretic security and the shannon limit are explained . in the section iii and iv
, we describe a theory of cppm . in the section
v and vi , we discuss on the security and implementation problem .
in the conventional cipher , the ciphertext @xmath3 is determined by the information bit @xmath4 and running key @xmath1 . this
is called non random cipher .
however , one can introduce more general cipher system so called random cipher by noise such that the ciphertext is defined as follows : @xmath5 where @xmath6 is noise .
such a random cipher by noise may provide a new property in the security . in shannon theory for the symmetric key cipher
, we have the following theorem .
+ + * theorem 1*(shannon , 1949 [ 7 ] ) + the information theoretic security against ciphertext only attack on data has the limit @xmath7 this is called shannon limit for the symmetric key cipher . to be beyond
the shannon limit is essential for fresh key generation by communication or information theoretic security against known plaintext attack in the symmetric key cipher . in the context of random cipher
, one can exceed this limit .
it is known that the necessary condition for exceeding the limit is @xmath8 [ 6,8,9 ] .
that is , the ciphertexts for bob and eve are different .
still the necessary and sufficient condition is not clear , but if the following relation holds , one can say that the cipher exceeds the shannon limit @xmath9 this means that eve can not pin down the information bit even if she gets a secret key after her measurement of the ciphertext while bob can do it . in the following sections
, we will clarify that cppm has indeed such a property .
the coherent pulse position modulation ( cppm ) cryptosystem has been proposed as a quantum cipher permitting asymptotical system parameters [ 1,6 ] .
alice encodes her classical messages by the block encoding where @xmath10-bit block @xmath11 ( @xmath12 ) corresponds to the pulse position modulation ( ppm ) quantum signals with @xmath13 slots , @xmath14 in addition , alice apply the unitary operator @xmath15 to ppm quantum signals @xmath16 , where the unitary operator @xmath15 is randomly chosen via running key @xmath17 generated by using prng on a secret key @xmath2 .
thus , alice gets cppm quantum signal states , @xmath18 which are sent to bob .
let us assume an ideal channel .
since the secret key @xmath2 , prng and the map @xmath19 are shared by alice and bob , bob can apply the unitary operator @xmath20 to the received cppm quantum signal @xmath21 and obtain the ppm quantum signal @xmath22 .
bob decodes the message by the direct detection for @xmath16 , which is known to be a suboptimal detection for ppm signals [ 10 ] .
then bob s block error rate is given by @xmath23 here @xmath24 holds for enough large signal energy @xmath25 .
in contrast , eve does not know the secret key @xmath2 and hence she must detect cppm quantum signals directly .
this makes eve s error probaility worse than bob s one .
in this section , we discuss a method for the construction of cppm quantum signals from ppm ones by the unitary operator associated with a symplectic transformation . the classical probability distribution @xmath26 is characterized by the _
characteristic function _
@xmath27\pi(dx)$ ] .
the characteristic function of gaussian distribution with the mean @xmath28 and the correlation matrix @xmath29 is given as @xmath30 $ ] .
we extend this argument to define the quantum gaussian state @xcite .
we consider self adjoint operators on a hilbert space @xmath31 , @xmath32 satisfying heisenberg ccr : @xmath33=i\delta _ { jk}\hbar i,\;\;[q_{j},q_{k}]=0,\;\;[p_{j},p_{k}]=0,\ ] ] where @xmath34 for @xmath35 and @xmath36 for @xmath37 .
let us introduce unitary operators @xmath38 for a real column @xmath39-vector @xmath40 and @xmath41^{t}.\ ] ] the operators @xmath42 satisfy the weyl - segal ccr @xmath43v(z+z^{\prime } ) , \label{weyl}\ ] ] where @xmath44 is the canonical symplectic form with @xmath45.\ ] ] the weyl - segal ccr is the rigorous counterpart of the heisenberg ccr , involving only bounded operators .
now we can define the quantum characteristic function as @xmath46 the transformation @xmath47 satisfying @xmath48 is called a _ symplectic transformation_.
we denote the totality of symplectic transformations by @xmath49 .
( [ s_cond ] ) can be rewritten as @xmath50 the symplectic transformation preserves weyl - seagl ccr ( [ weyl ] ) and hence it follows from stone - von neumann theorem that there exists the unitary operator @xmath51 satisfying [ 11 ] @xmath52 we call such derived operator @xmath51 the _ unitary operator associated with symplectic transformation _ @xmath47 .
the density operator @xmath53 is called gaussian if its quantum characteristic function has the form @xmath54.\ ] ] in an @xmath13-mode gaussian state , @xmath28 is a @xmath55-dimensional mean vector and @xmath56 is a @xmath57-corralation matrix .
the mean @xmath28 can be arbitrary vector ; the necessary and sufficient condition on the correlation matrix @xmath56 is given by @xmath58 the coherent state @xmath59 ( @xmath60 ) is the quantum gaussian state with the mean @xmath61 and the correlation matrix @xmath62,\ ] ] and the @xmath13-ary coherent state @xmath63 ( @xmath64 ) is the quantum gaussian state with the mean @xmath65 and the correlation matrix @xmath66.\ ] ] we study a way to generate cppm quantum signals by the unitary operator @xmath51 associated with a symplectic transformation .
any unitary operator composed of beam splitters and phase shifts can be described by a symplectic transformation .
first , let us consider the state @xmath67 for a general @xmath13-ary coherent state @xmath68 . by using eq .
( [ stone von neumann ] ) , the quantum characteristic function of the state @xmath67 is given as @xmath69 , \end{split}\ ] ] where @xmath28 and @xmath70 is the mean vector and the correlation matrix given by eqs .
( [ c ave ] ) and ( [ c cor ] ) respectively .
( [ cppm ch ] ) shows that the state @xmath67 is the quantum gaussian state with the mean @xmath71 and the correlation matrix @xmath72 .
our interest is devoted to the case where the state @xmath67 is an @xmath13-ary coherent state .
then the symplectic transformation @xmath73 should satisfy the condition @xmath74 , that is , @xmath75 where @xmath76 is the @xmath57 identity matrix and @xmath77 is the set of @xmath57 real matrices . denoting the totality of @xmath78-unitary matrices by @xmath79 , we have the relation @xmath80 here the matrix @xmath81 with real numbers @xmath82 and rotation matrices @xmath83 , corresponds to the matrix @xmath84 we can find that the unitary operator associated with @xmath85 transforms the state @xmath68 to the state @xmath86 , where @xmath87 and @xmath88 are related in the equation @xmath89 in particular , from the ppm quantum signals @xmath90 , @xmath91 , the cppm ones are generated as @xmath92 in other words , @xmath13-ary coherent states @xmath93 are the cppm quantum signals generated by applying the unitary operator associated with @xmath85 to the ppm quantum signals @xmath16 if and only if the matrix with @xmath94-elements @xmath95 is unitary .
we give a foundation for discussing security of cppm cryptosystem . without knowing
the secret key @xmath2 eve can not apply the appropriate unitary operator to cppm quantum signals , and hence she has to receive directly cppm quantum signals . since the quantum optimum receiver is unknown for such signals ,
we apply the heterodyne receiver , which is suboptimum and appropriate to discuss the performance of error .
this scheme is called heterodyne attack .
our main target in this subsection is to study the heterodyne attack on @xmath67 , where @xmath51 is the the unitary operator associated with @xmath85 , and @xmath96 is a general @xmath13-ary coherent state @xmath63 .
heterodyne detection is characterized by a family of operators with a parameter @xmath97 , @xmath98 the outcomes @xmath99 of the heterodyne detection for a coherent state @xmath100 appears with the probability density function @xmath101 which represents the normal distribution with the correlation matrix @xmath102 .
the outcomes @xmath103 of the indivisual heterodyne detection for @xmath67 obeys the probability density function , @xmath104 with @xmath105 . here , putting @xmath106 and taking account of eq ( [ rel alpha ] ) , we get @xmath107 note that @xmath108 represents the conjugate transpose and @xmath109 corresponds to the unitary operator @xmath110 . substituting this equation to eq .
( [ pdf ] ) , we obtain @xmath111 where @xmath112 is the probability density function with which the outcomes of heterodyne detection for the state @xmath96 appears . eq .
( [ pu ] ) shows that the vectors @xmath113 given by applying the unitary matrix @xmath109 to the outcomes @xmath114 obeys the probability density function @xmath112 .
it is difficult to evaluate the error performance for cppm quantum signals by heterodyne attack , because the randomness of prng has to be taken into account .
yuen showed the lower bound of the error performance by using heterodyne detection for the original ppm quantum signals [ 6 ] .
but it may be not tight one .
here we try another approach .
we allow eve to get the secret key @xmath2 after her measurement by heterodyne for cppm quantum signals and hence to know the unitary operator @xmath15 and the corresponding unitary matrix @xmath115 .
then , from the discussions in the subsection [ 7_1 ] , eve can apply the unitary matrix @xmath116 to obtain the vector @xmath117 , which obeys to the probability density function @xmath118 .
this fact enables us to apply the decoding process for ppm signals .
that is , eve may use maximum - likelihood decoding for @xmath117 , whose rule is to pick the @xmath11 for which @xmath119 is largest , and her error probability is given as follows [ 12 ] : @xmath120q_n(y)dy,\ ] ] where @xmath25 , and @xmath121^{n-1},\\ \phi(y)&=\frac{1}{\sqrt{2\pi}}\int _ { -\infty}^{y } \exp(-v^2/2)dv .
\end{split}\ ] ] we will compute the lower bounds of eve s error probability @xmath122 to evaluate its convergence speed .
the error probability @xmath123 is lower bounded as @xcite : @xmath124q_n(y)dy\\ \geq & q_n(z)\phi(z-\sqrt{2s } ) , \end{split}\ ] ] where the parameter @xmath40 can take any real number value .
putting @xmath125 and @xmath126 in ( [ kagen ] ) , we obtain @xmath127 let us consider the case of @xmath128 . then bob s error probability @xmath129 is less than @xmath130 , while @xmath123 converges to @xmath131 .
figure [ fig1 ] shows convergence behavior of lower bound for @xmath123 . in this figure ,
the lower bounds ( [ kagen ] ) for @xmath132 , are plotted with respect to @xmath133 . since the parameter @xmath134 in the lower bound ( [ kagen ] ) can take arbitrary real number , values of error probability @xmath123 exist the region above the graphs in figure [ fig1 ] .
note that the above values of @xmath134 are chosen as they give better lower bounds for @xmath123 . from figure @xmath135
, it is found that the convergence speed of lower bound for @xmath123 is very slow ; @xmath136 ( @xmath137 ) is needed to achieve the error probability 0.9 .
thus , in the cppm scheme , eve can not pin down the information bit even if she gets the true secret key @xmath2 and prng after her measurement , and consequently it has been proved that cppm satisfies eq(3 ) .
after her measurement for cppm quantum signals ]
according to the above analysis , one needs large number of @xmath10 when the signal energy is large . here
we give a requirement of channel bandwidth for the secure communication by cppm .
let us assume that the signal band width is @xmath138 when there is no coding . in our scheme , first one has to transform the @xmath10 input information bit sequence to ppm signal with @xmath139 slots .
second , such ppm signals are converted into cppm with the same number of slots .
if one wants to transmit such cppm signal with no delay , the required bandwidth is @xmath140 thus , the bandwidth exponentially increases with respect to @xmath10 .
since one needs the large @xmath10 to ensure the security , one needs a huge bandwidth . on the other hand
, we need to realize the unitary transformation to generate cppm quantum signals from ppm ones .
such transformations may be implemented by combination of beam splitters and phase shifts [ 6 ] , but to generate the cppm quantum signals with uniform distance for all signal , we need also large number of elements .
thus we need more detailed consideration for the practical use . in future work
, we will specify the realization method .
a crucial element of the coherent pulse position modulation cryptosystem is a generation of cppm quantum signals from ppm ones by a unitary operator . in this paper , we have given a proper theory for a unitary operator and a symplectic transformation basing on the quantum characteristic function .
furthermore , by using the above results , we have shown the lower bound of error probability in the case where eve gets the secret key after her measurement .
this result clearly guarantees the fresh key generation supported by the secret key encryption system .
we are grateful to dr .
usuda and the research staff in tamagawa university for the discussion on this subjects . | a quantum cipher supported by a secret key so called keyed communication in quantum noise ( kcq ) is very attractive in implementing high speed key generation and secure data transmission .
however , yuen-2000 as a basic model of kcq has a difficulty to ensure the quantitative security evaluation because all physical parameter for the cipher system is finite .
recently , an outline of a generalized scheme so called coherent pulse position modulation(cppm ) to show the rigorous security analysis is given , where the parameters are allowed to be asymptotical .
this may open a new way for the quantum key distribution with coherent states of considerable energy and high speed . in this paper , we clarify a generation method of cppm quantum signal by using a theory of unitary operator and symplectic transformation , and
show an asymptotic property of security and its numerical examples . |
in @xcite , guy david proposed to consider the plateau problem with sliding boundary conditions .
that is , given a closed set @xmath1 , and an initial closed set @xmath2 , we aim to find a competitor @xmath3 such that @xmath4 attains the infimum among all the competitors of @xmath5 , where @xmath6 is an integer between @xmath7 and @xmath8 .
the sliding conditions seem very natural to plateau s problem ( or soap films ) .
one of the advantages is that it may be easier to prove some regularity at the boundary .
in fact , paper @xcite paves the way to show the regularity . in the recent papers @xcite ,
the authors have proposed a direct approach to the plateau problem with sliding boundary conditions .
eventually they proved an existence result . which is that when @xmath9 is a closed set with @xmath10 , then there exists ( at least ) a sliding minimizer for the plateau problem with sliding boundary conditions .
the aim of the present paper is to study the regularity of the sliding minimizer . in @xcite ,
jean taylor proved that any @xmath11-dimensional reduced almost minimal set in @xmath0 is locally @xmath12-diffeomorphic to a minimal cone . here
we hope to prove a similar regularity result at boundary for sliding almost minimal set .
that is , at the boundary it is locally @xmath12-diffeomophic to a sliding minimal cone .
but unfortunately we do not prove the @xmath12-diffeomophic equivalence at this time . in this paper
, we show that if @xmath13 is a reduced @xmath11-dimensional set with a sliding boundary condition given by a smooth @xmath11-dimensional surface , then @xmath3 is locally bihlder equivalent to a sliding minimal cone . in @xcite and @xcite , guy
david has given a new , more detailed , proof of a good part of jean taylor s regularity theorem for almgren almost minimal sets of dimensional 2 in @xmath0 , and generalized it to @xmath14 .
at the same time , he proved a theorem of almost monotonicity of density for almost minimal sets away from the boundary .
in fact , his proof of hlder regularity relies on the theorem of almost monotonicity of density and a reifenberg parameterization . in @xcite
, he established a very similar result as in @xcite , a theorem of almost monotonicity of density at the boundary for sliding almost minimal sets .
this allow us to prove the hlder regularity of these sets on the boundary in some case . at the end of the paper
, we will discuss how to use regularity result blow to prove some existence results .
let us begin with some notation and definitions .
a gauge function is a nondecreasing function @xmath15\to[0,\infty]$ ] with @xmath16 .
let @xmath17 and an open set @xmath18 be given .
a @xmath19-deformation in @xmath20 is a family of maps @xmath21 from @xmath20 into itself such that @xmath22 the function @xmath23\times u\to u , ( t , x)\mapsto \varphi_{t}(x)\ ] ] is continuous , @xmath24 is relatively compact in @xmath20 and @xmath25 , where @xmath26}\left ( w_{t}\cup \varphi_t(w_t ) \right ) ,
\ w_t=\ { x\in u;\varphi_t(x)\neq x \}.\ ] ] we say that a relatively closed @xmath6-dimensional set @xmath27 is @xmath28-almost - minimal if it satisfies @xmath29 for any @xmath19-deformation @xmath21 . in @xcite , jean taylor proved that if @xmath20 is an open set in @xmath0 , @xmath3 is a reduced @xmath28-almost - minimal set and @xmath30 , then for any @xmath31 , there is a small neighborhood of @xmath32 contained in @xmath20 and in this neighborhood , @xmath3 is @xmath12 diffeomorphic to a @xmath11-dimensional minimal cone , while @xmath11-dimensional minimal cones are planes , cones of type @xmath33 and cones of type @xmath34 . in this paper
, we concentrate on boundary regularity , and always consider the following sliding boundary conditions .
let @xmath35 be a closed domain in @xmath14 .
let @xmath36 be a closed sets ( it will be consider as the sliding boundary ) .
let @xmath20 be an open set .
for @xmath17 , we say that a one parameter family @xmath21 of maps from @xmath20 into itself is a @xmath19-sliding - deformation in @xmath20 , if it satisfies the following properties : @xmath37 , @xmath38 is lipschitz , @xmath39 is continuous on @xmath40\times u$ ] , @xmath41 for any @xmath42 and any @xmath43 $ ] , @xmath24 is relatively compact in @xmath20 and @xmath25 .
let @xmath44 be closed in @xmath45 ; we say that a closed subset @xmath46 is a competitor of @xmath3 in @xmath20 , if @xmath47 for some sliding deformation @xmath48 in @xmath20 .
let @xmath49 be closed in @xmath20 .
we say that @xmath3 is @xmath28-sliding - almost - minimal , if for each @xmath17 and each @xmath19-sliding - deformation @xmath50 , we have @xmath51 where @xmath52 .
we say that @xmath3 is an @xmath53-sliding - almost - minimal set in @xmath20 if under the same circumstances , @xmath54 in the definition of @xmath28-sliding - almost - minimal set , we can replace inequality by the inequality @xmath55 at least if that @xmath36 is not too bed , see @xcite . when @xmath45 , @xmath36 and gauge function @xmath56 are clear , and @xmath57 . for simplicity
, we may say that an @xmath28-sliding - almost - minimal set @xmath3 is sliding almost minimal ( in @xmath45 with sliding boundary @xmath36 ) .
it quite easy to see that for any @xmath28-sliding - almost - minimal set @xmath3 , @xmath58 is @xmath59-almost - minimal .
we say that @xmath3 is ( sliding ) minimal in @xmath20 if it is ( sliding ) almost minimal with gauge function @xmath60 , that is , @xmath61 or @xmath62 for any ( sliding ) deformation @xmath21 in @xmath20 .
we say that a @xmath6-dimensional set @xmath3 is reduced if @xmath63 , where @xmath64 we can prove that @xmath65 and that @xmath66 is also ( sliding ) almost minimal when @xmath3 is ( sliding ) almost minimal , see for instance @xcite . in this paper
, we always assume that a sliding almost minimal set is reduced . for any set @xmath3 ,
any point @xmath67 and any radius @xmath68 , we set @xmath69 where @xmath70 denote the hausdorff measure of @xmath6-dimensional unity ball . if the limit @xmath71 exists , we will denote it by @xmath72 , and call it the density of @xmath3 at the point @xmath32 . when @xmath3 is given , and there is no danger of confusion , we may drop the subscript @xmath3 and denote it by @xmath73 .
a property of almost monotonicity of density for sliding almost minimal set will be often used .
that is , proposition 5.27 in @xcite and theorem 28.7 in @xcite .
we now put them together , it can be stated rough as follows : @xmath74 when @xmath75 small , where @xmath3 is a sliding almost minimal set with gauge function @xmath56 , and @xmath76 .
let s refer to @xcite and @xcite for more detailed statement .
we get that from that when @xmath3 is a sliding almost minimal set , for any @xmath67 , the density @xmath72 exists .
a blow - up limit of a set @xmath3 at @xmath67 is any closed set in @xmath14 that can be obtained as the limit of a sequence @xmath77 with @xmath78 . a set @xmath3 in @xmath14
is called a cone centered at origin @xmath7 if for any @xmath67 and any @xmath79 , @xmath80 . in general , a cone is a translation of a cone centered at origin .
suppose that @xmath3 is a sliding almost minimal set , and @xmath67 .
if @xmath32 is not contained in the sliding boundary , then any blow - up limit of @xmath3 at @xmath32 is a minimal cone in @xmath14 , see @xcite ; if @xmath32 is in the sliding boundary , then any blow - up limit of @xmath3 at @xmath32 is a sliding minimal cone , see @xcite .
we also refer to @xcite and @xcite for the basic properties of blow - up limits .
the main theorem of the paper is following : let @xmath81 be a connected closed set such that the boundary @xmath82 is a two - dimensional @xmath12 submanifold .
suppose that @xmath32 is a point in @xmath82 , @xmath20 is a neighborhood of @xmath32 , @xmath83 is an @xmath28-sliding - almost - minimal set with sliding boundary @xmath82 and @xmath84 .
then for each small @xmath85 , we can find a radius @xmath86 , a sliding minimal cone @xmath87 in @xmath45 with sliding boundary @xmath36 and a bihlder map @xmath88 such that @xmath89 where @xmath45 is a half space and @xmath36 is the boundary of @xmath45 .
the list of @xmath11-dimensional sliding minimal cones which contain @xmath36 is not complicated , which is following : @xmath36 , cones @xmath90 , where @xmath87 is a cone of type @xmath91 or @xmath92 .
see section [ se:2dimcones ] for precise definitions and see theorem [ thm : mccb ] for a precise statement .
it seems to be a reasonable condition for soap film that @xmath84 . in soap film experiments ,
if we dip a shape of wire into some soapy water , when we pull it out we shall obtain a surface created by the soap film .
the wire is considered as the sliding boundary , and the surface is consider as a sliding almost minimal set .
actually , this surface seems to contain the wire .
thus the assumption @xmath93 seems natural to the author
. it would be also very interesting to consider the regularity at the boundary of sliding almost minimal sets which do not necessarily contain the boundary .
but unfortunately , without the assumption @xmath94 , we do not have a satisfactory result . because in this case , the blow - up limits of @xmath3 at a point @xmath95 could be cones of type @xmath96 or cones of type @xmath97 .
see section [ se:2dimcones ] for a precise definitions of these cones .
when a blow - up limit is a cone of type @xmath97 , we will meet trouble ( figure [ fig : buv ] is an example of potential soap film for which regularity seems difficult to prove ) .
is a cone of type @xmath97,width=480 ]
in this section we discuss one dimensional sliding minimal sets in a half plane .
we discuss the one dimensional case , because it is very easy , and the list of one dimensional sliding cones will be used to classify the two dimensional sliding minimal cones in a half space . for simplicity , we assume that @xmath98 for any @xmath99 , and any @xmath100 , we set @xmath101 and @xmath102 it is very easy to see that the set @xmath103 and @xmath104 are sliding minimal .
it is also not hard to see that @xmath105 is minimal if and only if @xmath106 .
[ listofonedimensionalminimalcone ] let @xmath45 and @xmath36 be as in .
suppose that @xmath3 is a minimal cone in @xmath45 with sliding boundary @xmath36 centered at @xmath7 .
then @xmath3 is one of @xmath36 , @xmath107 , @xmath108 and @xmath109 for some @xmath110 .
let @xmath111 .
we note that @xmath112 is a finite set because otherwise @xmath113 .
write @xmath114 , and denote by @xmath115 the ray form @xmath7 through the point @xmath116 .
suppose @xmath117 , @xmath118 .
similarly to ( 10.3 ) in lemma 10.2 in @xcite , we can get that @xmath119 therefore , there are at most four point in @xmath112 .
case 1 , if there is only one point in @xmath112 , i.e. @xmath120 .
it is easy to see that @xmath121 and @xmath122 .
if @xmath123 , it is very easy to see that @xmath3 is minimal .
if @xmath124 , we put @xmath125 , then @xmath126 is a competitor of @xmath3 , and @xmath127
. then @xmath3 could not be minimal . in this case
, @xmath128 is a ray which is perpendicular to @xmath36 .
case 2 , there are two points in @xmath112 , i.e. @xmath129 . if @xmath130 and @xmath131 , then @xmath132 .
if @xmath130 and @xmath133 , then @xmath134\ ] ] is a competitor of @xmath3 and @xmath135 . then @xmath3 could not be minimal . if @xmath131 and @xmath122 , for the same reason as before , @xmath3 is not minimal .
if @xmath136 , we put @xmath137 , @xmath138 and @xmath139 with @xmath140 , then @xmath141\cap [ 0,a_2])\geq { \mathcal{h}}^1([a_1,\widetilde{a}_2]),\ ] ] and with equality if and only if @xmath142 .
it means that when @xmath143 , @xmath3 could not be minimal . we now suppose that @xmath144
. then @xmath145 , and @xmath146 because @xmath3 is minimal .
case 3 , there are three point in @xmath112 .
write @xmath147 , @xmath148 , @xmath149 and @xmath150 , @xmath151 . if @xmath152 and @xmath153 , then @xmath154 that is impossible .
if @xmath155 and @xmath153 , then @xmath156 thus @xmath157 . we can see that @xmath158\cup [ ( x_2,0),a_2]\cup [ 0,a_3]\ ] ] is a competitor of @xmath3 , and @xmath159 , thus @xmath3 could not be minimal .
similarly , we can see that we can not have @xmath160 and @xmath161 .
we now suppose that @xmath155 and @xmath162 , i.e. @xmath130 and @xmath163 .
if @xmath164 , then @xmath165 is a competitor of @xmath3 , and @xmath166 , @xmath3 is not minimal , impossible ! if @xmath167 , then @xmath168 .
that is , @xmath169 .
case 4 , there are four point in @xmath112 . if there are at least three point in @xmath170 , similarly to , that is impossible .
thus there at most two point in @xmath170 , so there are exactly two point in @xmath170 and exactly two point in @xmath171 .
we put @xmath172 , @xmath130 , @xmath149 , @xmath150 , @xmath173 , @xmath174 and @xmath175
. then @xmath176\cap [ 0,a_3 ] ) > { \mathcal{h}}^1([a_2,\widetilde{a}_2])+{\mathcal{h}}^1([a_3,\widetilde{a}_3]),\ ] ] which means that @xmath3 could not be minimal .
[ listofonedimensionalminimalset ] let @xmath45 and @xmath36 be as in .
suppose that @xmath3 is a sliding minimal set in @xmath45 with sliding boundary @xmath36 , and @xmath177 .
then either @xmath132 or @xmath178 for some @xmath179 . for @xmath68
, we put @xmath180 , @xmath181 and @xmath182 .
we claim that there exists a sequence @xmath183 , such that @xmath184 and there are at most three point in @xmath185 .
since @xmath3 is sliding minimal , we have that @xmath186 is nondecreasing and bounded , see ( * ? ? ?
* theorem 28.4 ) .
thus for any @xmath187 , we can find @xmath188 such that @xmath189 for @xmath190 , where we denote @xmath191 .
we can easily see that @xmath192 .
if we take @xmath193 , then @xmath194 we now let @xmath195 with @xmath196 and @xmath197 be as in proposition 30.3 in @xcite .
we take @xmath198 and apply proposition 30.3 in @xcite , and get that there is a minimal cone @xmath199 centered at @xmath7 such that @xmath200 if we put @xmath201 , then by lemma 8.10 in @xcite or theorem 3.2.22 in @xcite , @xmath202 combining this with , we can get that , for @xmath203 , @xmath204 but @xmath205 can be chosen arbitrarily large , thus we can find a sequence @xmath206 such that @xmath207 and @xmath208 . since @xmath209 , we have that @xmath210 for any @xmath211 , thus @xmath212 .
if @xmath213 for some @xmath211 , by minimality of @xmath3 , we can get that @xmath214 if @xmath215 for some @xmath211 , we suppose that @xmath216 if the points @xmath217 and @xmath7 are not in the same component of @xmath218 , then by minimality of @xmath3 , we can see that @xmath217 is the only point in the component of @xmath218 which contains the point @xmath217 , and @xmath214 if the points @xmath217 and @xmath7 are in the same component of @xmath218 , then there is a path in @xmath219 from @xmath217 to @xmath7 , we denote it by @xmath220\to e\cap \overline{b(0,s)}$ ] with @xmath221 and @xmath222 .
let @xmath223\mid \gamma(v)\in l_1 \}.\ ] ] then @xmath224\cup \left ( e\setminus b(0,s ) \right)\cup \left ( l_1\cap b(0,s ) \right)\ ] ] and @xmath225\cup \left ( e\setminus b(0,s ) \right)\cup \left ( l_1\cap b(0,s ) \right)\ ] ] are competitors of @xmath3 , and @xmath226 by minimality of @xmath3 , we get that @xmath227 .
we put @xmath228 , then @xmath229 if there exists a sequence @xmath230 such that @xmath231 , then @xmath232 for any @xmath233 ; thus we get that @xmath132 . if there exist an integer @xmath234 such that @xmath235 for @xmath236 , then @xmath237 and @xmath238 for any @xmath236 . by putting @xmath239
, we get that @xmath178 .
in this section we consider a simple case in @xmath0 : our domain @xmath45 is a half space , and the boundary @xmath36 is the plane which is the boundary of @xmath45 . in the domain @xmath45
, we will see what does a sliding minimal cone look like . for simplicity , we assume that @xmath240 let us refer to paper @xcite for the definition of cones of type @xmath33 and @xmath34 .
we say that a cone @xmath241 is of type @xmath91 , if @xmath87 is a closed half plane which is perpendicular to @xmath36 and through @xmath7 , i.e. the intersection of @xmath45 with a plane which is through @xmath7 and meets @xmath36 perpendicularly ; similarly we say that a cone @xmath241 is of type @xmath92 if it is a intersection of @xmath45 with a cone in @xmath0 of type @xmath33which is perpendicular to @xmath36 . recall that a cone @xmath87 in @xmath0 of type @xmath33 is the union of there half planes bounded by a line @xmath242 , called the spine of @xmath87 . here
we say that a cone of type @xmath33 is perpendicular to @xmath36 , if the spine of the cone is perpendicular to @xmath36 .
we will check that cones of type @xmath91 or @xmath92 are sliding minimal .
let @xmath87 be a cone of type @xmath34 , we say that @xmath87 is perpendicular to @xmath36 if the center of @xmath87 locates at the origin and @xmath243 is a cone of type @xmath92 in the domain @xmath244 .
we say that a cone @xmath241 is of type @xmath96 if it is the intersection of @xmath45 with a cone of type @xmath34 which is perpendicular to @xmath36 . in this paper
, we do not discuss whether or not a cone of type @xmath96 is sliding minimal .
a cone @xmath245 is called of type @xmath97 is it can be written as @xmath246 where @xmath247 is a rotation which maps @xmath36 into @xmath36 , @xmath109 is cone in a half plane defined as in section [ se:1dimcones ] and @xmath248 . [
le : plmc ] let @xmath45 , @xmath249 be as in . if @xmath87 is a cone of type @xmath91 or @xmath92 , then @xmath87 is a sliding minimal cone .
if @xmath250 and @xmath251 is a sliding minimal cone of type @xmath91 or @xmath92 , then @xmath87 is also a sliding minimal cone .
suppose that @xmath87 is of type @xmath91 or @xmath92 , which is not sliding minimal .
then there is a competitor of @xmath87 , say @xmath3 , such that @xmath252 let @xmath253 be the reflection with respect to the plane @xmath36 .
that is , for any @xmath254 , @xmath255 .
then @xmath256 is a competitor of @xmath257 , and @xmath258 but we know that @xmath259 is a plane or a cone of type @xmath33 , which is minimal in @xmath0 , that gives a contradiction .
now suppose that @xmath250 , where @xmath251 is cone of type @xmath91 or @xmath92 .
let @xmath3 be any competitor of @xmath87 .
suppose that @xmath3 coincide with @xmath87 out of the ball @xmath260 .
let @xmath261 be the function defined by @xmath262 . by using lemma 8.10 in @xcite or theorem 3.2.22 in @xcite
, we get that @xmath263 where @xmath264 is the approximate jacobian , see @xcite .
we can check that @xmath265 for any @xmath266 .
thus @xmath267 for any @xmath268 , @xmath269 is a plane , and @xmath270 is a line or a @xmath271 in this plane , so it is minimal in the plane . but @xmath272 coincide with @xmath270 out of the ball @xmath273 , and it is not hard to check that @xmath274 is connected .
thus @xmath275 hence @xmath276 since @xmath277 , and @xmath251 is a cone of type @xmath91 or @xmath92 , we have that @xmath278 we get that @xmath279 thus @xmath280 and @xmath87 is minimal .
let @xmath281 be any convex polyheddron , @xmath32 be a point in the interior of @xmath281 .
if @xmath282 is a compact set with @xmath283 , then we can find a lipschitz map @xmath284 such that @xmath285 indeed , we take a very small ball @xmath286 such that @xmath287 , and consider the map @xmath288 defined by @xmath289 @xmath290 is lipschitz on @xmath291 . by the kirszbraun s theorem @xcite , we can find a lipschitz map @xmath292 such that @xmath293 let @xmath45 , @xmath249 be as in . if @xmath251 is a cone of type @xmath96 , then the cone @xmath294 is not minimal .
here we do not want to talk about whether or not a cone of type @xmath96 is minimal , it is not so obvious . recall that a cone of type @xmath34 has six faces , that meet by sets of three and with @xmath295 angles along four edges ( half lines emanating from the center ) .
we put @xmath296 , @xmath297 , @xmath298 , @xmath299 , @xmath300 , @xmath301 , @xmath302 .
we denote by @xmath303 the triangular prism @xmath304 , by @xmath305 the union of eight edges of @xmath303 . without loss of generality
, we assume that @xmath306 we denote by @xmath307 , @xmath308 , @xmath309 and @xmath310 the faces @xmath311 , @xmath312 , @xmath313 and @xmath314 of the prism @xmath303 respectively .
consider @xmath315 .
we will show that @xmath316 is a competitor of @xmath87 .
we take @xmath317 , then @xmath318 is in the interior of the triangular prism .
we take a lipschitz map @xmath319 as in .
then @xmath320 is a competitor of @xmath87 .
since @xmath321 consists of faces ( triangles ) @xmath311 , @xmath322 , @xmath323 , @xmath324 , @xmath325 , @xmath326 and @xmath327 . by a simple calculation
, we can get that @xmath328 similarly , @xmath329 consists of faces @xmath307 , @xmath308 , @xmath309 and @xmath310 , thus @xmath330 therefore @xmath331 and @xmath87 is not minimal
. let @xmath332 be a closed subset of @xmath14 .
let @xmath333 be given .
we say that a nonempty bounded subset @xmath334 is @xmath6-dimensional @xmath335-minimal relative to @xmath332 if @xmath336 and @xmath337 whenever @xmath338 is lipschitz with @xmath339 and @xmath340 , where @xmath341 . when @xmath342 , @xmath343 , @xmath44 is a reduced bounded @xmath28-@xmath53-sliding - almost - minimal set , then it is very easy to see that @xmath3 is also @xmath335-minimal relative to @xmath332 . with help of this property ,
we can use a result of morgan ( * ? ? ?
* regularity theorem 3.8 ) , which is stated as follows .
[ thm : regularityonsphere ] fix @xmath333 .
let @xmath332 be a closed subset of @xmath14 .
let @xmath344 be a one - dimensional @xmath335-minimal set with respect to @xmath332 .
then @xmath344 consists of @xmath345 curves that can only meet in three at isolated points of @xmath346 and with @xmath347 angles .
let @xmath3 be a sliding minimal cone in @xmath45 .
set @xmath348 , @xmath349 .
we want to show that @xmath344 is @xmath350-minimal with respect to @xmath342 for some @xmath351 .
[ prop : regularityonshpere ] let @xmath352 be as in , @xmath342 .
let @xmath3 be a reduced sliding minimal cone in @xmath45 , @xmath353 . if @xmath354 , then @xmath112 is @xmath53-sliding - almost - minimal for some gauge function @xmath56 such that @xmath355 for @xmath356 .
let @xmath50 be a deformation with @xmath357 , where @xmath24 as in .
if @xmath358 , we have nothing to prove .
we now suppose that @xmath359 ; we can find a point @xmath360 , such that @xmath361 .
we consider the lipschitz function @xmath362 $ ] defined by @xmath363 we consider @xmath364 defined by @xmath365 when @xmath366 or @xmath367 , @xmath368 ; when @xmath369 , @xmath370 .
also @xmath371 is a lipschitz map with @xmath372 we put @xmath373 ; then @xmath374 for @xmath187 small , we consider the lipschitz map @xmath375 defined by @xmath376 where @xmath377 $ ] given by @xmath378 it is clear that @xmath379 for @xmath380 , @xmath381 for @xmath382 .
we consider the map @xmath383 given by @xmath384 it is lipschitz , thus the map @xmath385 is also lipschitz .
it is easy to see that @xmath386 is a competitor of @xmath3 , because that @xmath387 defined by @xmath388 is a deformation .
we will compare @xmath389 with @xmath3 .
since @xmath3 is a cone , @xmath390 lie in the line through @xmath7 and @xmath32 , @xmath391 is the radial projection into the sphere @xmath392 on the annulus @xmath393 , and @xmath394 is identity out of the ball @xmath395 , we can get that @xmath389 and @xmath3 coincide out of the ball @xmath396 . since @xmath3 is minimal , we have that @xmath397 recall that @xmath361 ; if we put @xmath398 where @xmath399 , then @xmath400 , thus on the sphere @xmath392 , we can see that @xmath389 and @xmath3 coincide out of @xmath401 , thus we can easily get that @xmath402 applying theorem 3.2.22 in @xcite , we have @xmath403 by the construction of @xmath389 , we know that @xmath389 coincide with a cone in the ball @xmath273 , thus the same reason as above , we have @xmath404 we combine this with , and , and get that @xmath405 by our construction of @xmath406 , we have that for any @xmath407 , if @xmath408 , i.e. , @xmath409 , then @xmath410 , thus @xmath411 .
we get that @xmath412 since @xmath413 , we can show that @xmath414 combine this with , and get that @xmath415 and by and , @xmath416 the result immediately follows .
[ prop : onsphere ] let @xmath45 , @xmath249 be as in .
let @xmath44 be a minimal cone , and set @xmath111 .
then @xmath112 consists of arcs @xmath417 of great circles .
these arcs can only meet at their extremities .
for each extremity , if it is not in @xmath36 , then it is a common extremity of exactly three arcs which meet with @xmath295 angles . a point in @xmath418
is called to be a @xmath271-point if it is a common extremity of exactly three curves which meet with @xmath295 angles .
applying theorem [ thm : regularityonsphere ] and proposition [ prop : regularityonshpere ] , we can get @xmath112 consists of @xmath419 curves , these curves only meet at their extremities . for each extremity , if it is not in @xmath36 , then it is a common extremity of exactly three curves which meet with @xmath295 . for any point @xmath32 in the interior of such a curves @xmath420 , by the same proof as in (
* proposition 14.1 ) , we can get that there is a neighborhood @xmath421 such that @xmath422 is an arc of great circles . from this
, we can immediately deduce the result .
[ le : bpc ] let @xmath423 be as in the proposition above .
for any @xmath42 , we denote by @xmath424 the half plane through @xmath7 which is perpendicular to @xmath36 and the straight line joining @xmath32 and @xmath7 .
then , for any point @xmath425 , any blow - up limit of @xmath112 at @xmath32 is a sliding minimal cone in @xmath424 with sliding boundary @xmath426 .
the proof of this lemma is almost the same as in the first part of the proof of theorem 8.23 in @xcite . without loss of generality
, we assume that @xmath427 .
then @xmath428 , @xmath429 .
let @xmath430 , @xmath431 .
suppose that @xmath432 and @xmath433 it is quite easy to see that @xmath434 and @xmath435 .
theorem 24.13 in @xcite says that @xmath436 is a sliding minimal cone in @xmath45 with sliding boundary @xmath36 .
we denote by @xmath437 the line though the points @xmath7 and @xmath32 . as in @xcite , page 140
, we can get that @xmath438 similarly , we can get @xmath439 is a sliding minimal cone in @xmath424 with sliding boundary @xmath440 .
let us check that @xmath441 .
it suffices to show that @xmath442 , since we already know that @xmath443 .
we take any @xmath444 , then there exists a sequence @xmath445 such that @xmath446 since @xmath447 , and @xmath424 is perpendicular to the line @xmath437 which pass through the points @xmath7 and @xmath32 , we get that the angles between the line @xmath437 and the segments which join the pints @xmath32 and @xmath448 tend to @xmath449 , i.e. @xmath450 if @xmath451 , by , we get that @xmath452 if @xmath453 , we will show that @xmath454 we put @xmath455 .
then @xmath456 . since @xmath453 , we have that @xmath457 and @xmath458 for @xmath459 large .
we consider the triangle formed by the vertices @xmath7 , @xmath32 and @xmath448 .
we get that @xmath460 .
thus @xmath461 since @xmath462 and @xmath463 we get that @xmath464 hence @xmath465 in the case @xmath453 , we get , from and , that @xmath466 thus , from and , we get that @xmath467 but we know that @xmath468 , thus @xmath469 .
[ lemma : boundaryofcone ] let @xmath470 be as in the proposition above , @xmath349 .
for any @xmath471 , there is a radius @xmath68 such that there is no @xmath271-point in @xmath472 . moreover , if a blow - up limit of @xmath112 at @xmath32 is a cone @xmath473 for some @xmath474 $ ] , then @xmath475 is a union of two arcs of great circles meeting at @xmath32 ; in the other cases , @xmath472 is an arc of great circle which perpendicular to @xmath36 .
we will prove that there ais only finite number of @xmath271 points in @xmath344 .
we denote @xmath476 .
let @xmath477 be a connected component of @xmath478 , @xmath479 be the closure of @xmath477 if @xmath480 , then @xmath479 is a convex .
indeed , we get , from proposition [ prop : onsphere ] , that each @xmath271-point in @xmath344 must connect three arcs , these three arcs meet with @xmath295 .
thus at each corner of @xmath481 , the interior angle of @xmath481 at this point must have be @xmath295 , and @xmath479 must be convex .
now , if @xmath482 , @xmath479 is also convex .
for the same reason , if the vertex of a corner of @xmath481 is contained in @xmath483 , then the interior angle of @xmath481 at this point must have be @xmath295 .
if the vertex of a corner of @xmath481 is contained in @xmath36 , then the the interior angle of @xmath481 at this point is no more than @xmath484 .
thus @xmath479 must be convex .
the number corners in @xmath481 must be finite .
indeed , there are at most four corners which touch the boundary @xmath36 , because @xmath477 is convex .
if there are infintely many corners in @xmath481 , then we can very easily to find @xmath485 corners , saying at points @xmath486 , such that these @xmath485 points are contained in @xmath344 , and the geodesic connecting @xmath487 and @xmath488 is contained in @xmath481 , @xmath489 .
we now consider the convex spherical polygon @xmath490 . by using the gauss - bonnet theorem , for example see (
* theorem .2.7 ) , we get that @xmath491 where @xmath492 and @xmath493 are the exterior angle of the corners of @xmath494 at point @xmath495 and @xmath496 respectivly .
but that is impossible , the equation gives an absurdity .
if @xmath479 are contained in @xmath483 , we assume that @xmath481 has @xmath8 cornes , then gauss - bonnet theorem says that @xmath497 thus @xmath498 , and @xmath499 . since the totall area of @xmath500 is @xmath371 ,
there are at most @xmath501 such connected components .
thus there is only a finite number of @xmath271-point in @xmath344 ; otherwise , it should be infintely many connected component of @xmath478 such that its corners does not touch @xmath36 . since there is only a finite number of @xmath271-point in @xmath344
, we get that for any @xmath471 , there is a radius @xmath502 such that there is no @xmath271-point in @xmath503 .
since @xmath112 is sliding almost minimal , any blow - up limit of @xmath112 at @xmath32 is a sliding minimal cone , denote by @xmath87 , and @xmath504 if @xmath87 is a cone like @xmath473 for some @xmath474 $ ] , then @xmath505 must be two arcs , each of these two arcs is a part of a great cicle , and these two arcs meet at @xmath32 with angle @xmath506 .
if @xmath87 is a half line perpendicular to @xmath36 , then @xmath505 is an arc which is a part of a great cicle , perpendicular to @xmath36 and through @xmath32 .
if @xmath87 is the union of a line in @xmath36 and a half line which is perpendicular to @xmath36 , then @xmath112 is the union of @xmath507 and an arc which is a part of a great cicle , perpendicular to @xmath36 and through @xmath32 .
[ le : ppc ] let @xmath45 , @xmath36 be as in .
let @xmath44 be a sliding minimal cone , @xmath508 , @xmath349 .
suppose that for each @xmath471 , there is a radius @xmath68 such that @xmath509 is an arc of a great circle which is perpendicular to @xmath36 .
then there are only there possible kinds of @xmath344 , that is , @xmath510 , where @xmath87 is a sliding minimal cone of type of one of @xmath91 , @xmath92 and @xmath96 . and hence , @xmath511 or @xmath512 where @xmath251 is a cone of type @xmath91 or @xmath92 .
we put @xmath513 .
let @xmath477 be a connected component of @xmath514 , @xmath479 be the closure of @xmath477 . by proposition [ prop : onsphere ] and lemma [ lemma : boundaryofcone ] ,
the boundary of @xmath479 is a spherical polygon whose sides are geodesics of the unit sphere . using the gauss - bonnet theorem ,
see ( * ? ? ?
* theorem .2.7 ) , we get that @xmath515 where @xmath516 are the exterior angle of the corners of @xmath494 . from lemma [ lemma : boundaryofcone ] and proposition [
prop : onsphere ] , we can see that , if a corner touch @xmath36 , then , in the situation of this lemma , the corresponding exterior angle must be @xmath449 ; if a corner do not touch @xmath36 , the corresponding exterior angle must be always @xmath517 .
it is quite clear that there are at least two corners on @xmath494 , and it can not happen that there only one corner touching @xmath36 .
we now consider the equation . if @xmath518 , then the two corners must touch @xmath36 , thus @xmath519 , and @xmath520 .
@xmath479 is a quarter of unity sphere .
let us split @xmath521 into two cases .
if there is no corner touching @xmath36 , then @xmath522 , and @xmath479 is an equilateral polar triangle with @xmath523 .
if it has corner on @xmath494 , then there are at least two , thus @xmath519 , @xmath524 , and @xmath479 is a isosceles polar triangle with @xmath525 .
more precisely , the base of @xmath479 is an arc contained in @xmath526 with length @xmath527 , the vertex opposite to the base is the point @xmath528 .
similarly , for @xmath529 , we can get two kinds of spherical quadrilaterals , one is spherical quadrilaterals with equal angles @xmath527 and with area @xmath527 , another one is spherical quadrilaterals with one side contained in @xmath526 and with area @xmath517
. we can easily see that @xmath8 can not be larger than @xmath530 ; otherwise , we can deduce from that @xmath531 , which is impossible .
for the same reason , when @xmath532 , there is only one kind of spherical pentagons .
that is , a spherical with all of corners are contained in @xmath483 and with area @xmath517
. since each connected component of @xmath533 has at least area @xmath517 , and the total area of @xmath534 is @xmath535 , there are at most six connected component .
if there is no @xmath271 point on @xmath483 , @xmath536 must be a half circle which is contained in @xmath500 and perpendicular to @xmath36 .
thus @xmath511 or @xmath537 , where @xmath87 is a cone of type @xmath91 , hence @xmath510 .
if there is only one @xmath271 point on @xmath483 , then each connect component of @xmath533 must be a polar triangle with base contained in @xmath526 . by our discussion for @xmath521
, we get that each such connected component is an isosceles polar triangle with area @xmath527 .
thus this @xmath271 point must be @xmath528 , and @xmath511 or @xmath537 , where @xmath87 is a cone of type @xmath92 , hence @xmath510
. if there are two @xmath271 points on @xmath483 , then there are at least two polar triangles with base contained in @xmath526 , and the vertices opposite to the bases must be the point @xmath528 , that is impossible
. if there are three @xmath271 points on @xmath483 , then these three points must be the vertices of a polar triangle which is contained in @xmath483 , and this triangle is an equilateral polar triangle with area @xmath371 .
each side of this triangle is a side of spherical quadrilateral , and the opposite side ot this spherical quadrilateral is contained in @xmath526 . in this case @xmath511 or @xmath537 , where @xmath87 is a cone of type @xmath96 , thus @xmath510 . since each polar triangle contained in @xmath483 has area @xmath371 , there only one such triangle .
if there are four @xmath271 point in @xmath483 , then these four point in @xmath536 must form a spherical quadrilateral , and @xmath514 consists of five region , that is , five connect component , each of them is a spherical quadrilateral , one of them is contained in @xmath483 , and each of the rest quadrilateral must has one side contained in @xmath36 .
without loss of generality , we suppose that @xmath538 , then those four @xmath271 points are @xmath539 , @xmath540 , @xmath541 and @xmath542 , we denote them by @xmath543 , @xmath544 , @xmath545 and @xmath546 respectively . in this case , we will show that is impossible .
we put @xmath547 , @xmath548 , @xmath549 and @xmath550 , and denote by @xmath303 the cube @xmath551 , by @xmath305 the union of the edges of cube @xmath303 .
we denote by @xmath307 the face @xmath552 of the cube @xmath303 , denote by @xmath308 and @xmath309 the rectangle @xmath553 and @xmath554 respectively .
[ fig : polytope ] we denote by @xmath555 and @xmath556 the points @xmath557 and @xmath558 respectively , and denote by @xmath559 , @xmath559 , @xmath560 and @xmath561 the polyhedrons @xmath562 , @xmath563 , @xmath564 and @xmath565 respectively .
we now take points @xmath566 , @xmath567 , @xmath568 and @xmath569 in the interior of pyramids @xmath570 , @xmath571 , @xmath572 and @xmath573 respectively .
let maps @xmath574 , @xmath575 , @xmath576 and @xmath577 be as in .
we take @xmath578 .
then @xmath579 is a lipschitz map .
we now consider the competitor @xmath580 of @xmath3 .
we denote @xmath581 . by a simple calculation
, we can get that @xmath582 that is contradict to minimality of @xmath3 .
it could not happen that there are at least six @xmath271 point in @xmath483 , because otherwise , there will be at least six spherical quadrilaterals which touch @xmath36 , but we know that such a quadrilateral has area @xmath517 , and total area of @xmath483 is @xmath535 , that is impossible .
if there are five @xmath271 point in @xmath483 , similar to above case , these five point form a spherical pentagon . by the same techniques used for the above case , we can prove that this is impossible .
[ le : ppcv ] let @xmath583 be as in .
let @xmath44 be a sliding minimal cone , @xmath508 .
if there exists a point @xmath584 such that a blow - up limit of @xmath112 at @xmath318 is a sliding minimal cone @xmath109 for some @xmath585 $ ] , then @xmath3 is a cone of type @xmath97 .
by lemma [ lemma : boundaryofcone ] , there exists a radius @xmath68 such that @xmath586 is a union of two arcs .
that is , @xmath587 where @xmath87 is a cone of type @xmath97 .
without loss of generality , we assume @xmath588 . we denote @xmath513 .
let @xmath477 be a connected component of @xmath514 which contains a corner with interior angle @xmath589 .
using the gauss - bonnet theorem , see ( * ? ? ?
* theorem .2.7 ) , we get that @xmath590 where @xmath516 are the exterior angle of the corners of @xmath494 .
we know that there are two corners which touch the boundary @xmath36 , assume that @xmath591 and @xmath493 are the corresponding exterior angles .
it is quite easy to see that @xmath477 is contained in a spherical lune enclosed by two great circles with angle @xmath589 , thus @xmath592 .
if @xmath593 , then @xmath594 , @xmath595 thus @xmath596 since @xmath597 $ ] , we get that @xmath598 , that is impposible . if @xmath599 , then @xmath600 , thus @xmath601 since @xmath602 , we get that @xmath603 , hence @xmath518 . in this case , @xmath477 must be a spherical lune enclosed by two great circles with angle @xmath589 , and @xmath604 .
[ thm : mccb ] let @xmath45 , @xmath249 be as in .
let @xmath44 be a sliding minimal cone .
if @xmath209 and @xmath605 .
then @xmath606 , @xmath87 is a cone of type @xmath91 or @xmath92 .
the result immediately follows from lemma [ lemma : boundaryofcone ] , lemma [ le : ppc ] and lemma [ le : ppcv ] .
indeed , by putting @xmath607 and @xmath349 , lemma [ le : bpc ] says that any blow - up limit of @xmath112 at a point @xmath608 is a one dimensional sliding minimal cone . by lemma [ listofonedimensionalminimalcone ] , there exist only there possible cases for such a minimal cone .
that is , a half line @xmath107 perpendicular to @xmath36 , or @xmath107 union the line which pass though @xmath32 perpendicular the segment @xmath609 $ ] and is contained in @xmath36 , or a cone @xmath109 . if it is a cone @xmath109 , then by lemma [ le : ppcv ] , @xmath604 , which is impossible . for any @xmath471 , by lemma [ lemma : boundaryofcone ]
, there exists a radius @xmath68 such that @xmath472 is an arc of a great cicle which is perpendicular to @xmath36 , and by lemma [ le : ppc ] , @xmath606 , where @xmath87 is a cone of type of one of @xmath91 , @xmath92 and @xmath96 , but for the last case , it is impossible , because we know that @xmath610 is not minimal when @xmath87 is of type @xmath96 . we get that @xmath606 , where @xmath87 is a cone of type @xmath91 or @xmath92 . [ re : listofminimalcones ] we claim that the list of sliding minimal cones is the following : cones of type @xmath91 , cones of type @xmath92 , the plane @xmath36 , cones like @xmath611 with @xmath612 , cones @xmath90 where @xmath87 are cones of type @xmath91 or @xmath92 , and cones type @xmath96 .
we did not prove that a cone of type @xmath96 is sliding minimal .
indeed it can probably be proved by using calibration , but this may take us too much time , we do not want to do it here .
it is also not too hard to check that a cone like @xmath613 is sliding minimal if and only if @xmath612 .
one of possible ways to do this is to sue almost the same technique as in lemma [ le : plmc ] , but again we omit it .
in fact , we do not need know whether or not a cone of type @xmath96 or like @xmath613 is minimal in this paper . for the rest in the list , we know from lemma [ lemma : boundaryofcone ] ,
lemma [ le : ppc ] and lemma [ le : ppcv ] that they are sliding minimal .
we want to use a result , theorem 2.2 in @xcite . but here we are in the half space , the theorem can not be used directly , it should be adapted a little bit .
let @xmath8 and @xmath6 be two integers with @xmath614 .
we take @xmath615 we let @xmath616 be the reflection with respect to @xmath36 .
that is , the function @xmath617 defined by @xmath618 let @xmath619 be a class of sets defined as in @xcite , which consists of 3 kinds of cones ( centered at any point in @xmath14 ) of dimension @xmath6 in @xmath14 .
in particular , if @xmath521 , @xmath620 , @xmath619 consists of planes , cones which are the union of three half planes bounded by a line while the angle between any two half planes is larger than a constant @xmath621 ( they look like cones of type @xmath33 ) , and cones which are union of several faces that meet only by sets of three and with angles between two adjacent faces and angles between the spines larger than a constant @xmath621 ( cones of type @xmath34 and cones look like of type @xmath34 are such cones ; cone @xmath622 is also a such cone , where @xmath87 is a cone of type @xmath92 or @xmath96 .
of course , far more than these cones ) . for any @xmath623 , @xmath68
, we will consider @xmath624 a variant of the hausdorff distance on closed sets , which is defined by @xmath625 if @xmath3 and @xmath436 are two cones centered at @xmath32 , then @xmath626 for any @xmath68 .
let @xmath627 be a compact set that contains origin with @xmath628 , and suppose that for each @xmath629 and each ball @xmath630 , we can find @xmath631 that contains @xmath32 , such that @xmath632 suppose , additionally , that @xmath633 .
if @xmath187 is small enough , depending only on @xmath8 , @xmath6 and @xmath634 , then there exist a cone @xmath635 centered at origin and a mapping @xmath636 with the following properties : @xmath637 where @xmath589 only depends on only @xmath8 , @xmath6 , @xmath634 and @xmath197 , and @xmath634 is defined as in ( * ? ? ?
* ( 2.7 ) and ( 2,8 ) ) .
the proof is essential the same as in @xcite , we only need do a little change .
here we use same notation as in @xcite .
we firstly remark that @xmath638 , @xmath639 or @xmath640 , where @xmath641 are defined as in ( * ? ? ?
* , p.12 ) .
next , we modify a little the construction of a good covering of @xmath3 at scale @xmath642 , that is in section 5 in ( * ? ? ?
* covering and partition of unity ) .
the first step is just same the as the original construction ; if the condition ( 4.36 ) in @xcite holds , we cover @xmath643 with the ball @xmath644 , and set @xmath645 ; if the condition ( 4.35 ) in @xcite holds , we take @xmath646 and choose no ball . in the second step , for the construction of a covering of @xmath647
we modify a little the original construction to adapt to our case .
we put @xmath648 , then @xmath649 .
select a maximal subset @xmath650 of @xmath651 , with the property that different points of @xmath650 have distances at least @xmath652 .
we put @xmath653 , and for accounting reasons , we suppose that @xmath654 , @xmath655 , and that @xmath656 , @xmath657 .
let @xmath658 , @xmath659 .
we put @xmath660 and @xmath661 for @xmath662 .
we can see that the balls @xmath663 , @xmath662 , cover @xmath664 . in the third step
, we put @xmath665 similarly to the above step , put @xmath666 , and select a maximal subset @xmath667 of @xmath668 , with the property that different points of @xmath667 have distances at least @xmath669 , and then suppose that @xmath670 with @xmath671 , and that @xmath672 with @xmath673 .
set @xmath674 , and @xmath675 for @xmath676 .
it is very easy to see that the balls @xmath663 , @xmath676 , cover @xmath677 . for the fourth and last step of the construction of the covering , we put @xmath678 we put @xmath679 , and pick a maximal subset @xmath680 of @xmath681 , such that different points of @xmath680 have distance at least @xmath682 , and then suppose that @xmath683 with @xmath684 , and that @xmath685 with @xmath686 .
set @xmath687 , and @xmath688 for @xmath689 , then the balls @xmath663 , @xmath689 , cover @xmath690 . for the selection of a partition of unity in equation ( 5.10 ) in @xcite ,
we choose the @xmath691 as the translation and dilation of a same model @xmath692 , where @xmath692 is a smooth function such that @xmath693 in @xmath694 , @xmath695 out of @xmath696 , @xmath697 everywhere , and @xmath698
. the rest of proof will be the same as in @xcite .
we now verify that @xmath699 .
it is clear that @xmath700 , @xmath701 , @xmath702 , and @xmath703 . by our construction of @xmath704 , @xmath705 and @xmath706
, we can see that @xmath707 thus @xmath708 . by induction on @xmath8
, we can get that @xmath709 for all @xmath710 .
@xmath711 is the limit of the sequence @xmath712 , thus @xmath713 . finally , by the same proof as above
, we can prove that @xmath714 .
[ co : pt ] for each small @xmath85 , we can find @xmath187 , that depends only on @xmath8 , @xmath195 and @xmath634 such that if @xmath44 is a closed set , @xmath715 and @xmath68 are such that for @xmath716 and @xmath717 , we can find @xmath718 , which is a minimal cone in @xmath0 when @xmath719 and a sliding minimal cone in @xmath45 with boundary @xmath36 when @xmath720 , such that @xmath721 and in addition @xmath722 is sliding minimal cone centered at @xmath7 , then there is a sliding minimal cone centered at origin and a mapping @xmath723 with the following properties : @xmath724
in this section , we restrict ourselves to the half space @xmath45 , and prove some boundary regularity for sliding almost minimal sets . [ mainlemma ]
let @xmath45 and @xmath36 be as in , @xmath20 an open set .
suppose that @xmath49 is @xmath28-sliding - almost - minimal .
for each @xmath85 , we can find @xmath725 such that if @xmath726 and @xmath68 are such that @xmath727 then for every @xmath728 $ ] there is a sliding minimal cone @xmath729 centered at @xmath32 , such that @xmath730 and for any ball @xmath731 , @xmath732 moreover , if @xmath209 , then we can suppose that @xmath733 .
this lemma is directly following from proposition 30.19 in @xcite .
if @xmath209 , by the original proof in , we can go further and assert that @xmath734 , the proof will be same , we do not even need to do any extra effort .
if @xmath3 is a sliding almost minimal set , then for any @xmath735 , any blow - up limit of @xmath3 at @xmath32 is a sliding minimal cone , see ( * ? ? ?
* theorem 24.13 ) . moreover , the density of any blow - up limit at origin is aways the value @xmath72 , see proposition 7.31 @xcite and corollary 29.53 in @xcite . by remark [ re : listofminimalcones ] , @xmath736 where we denote by @xmath737 the density of cones of type @xmath96 at origin .
in fact , @xmath738 . if @xmath739 , then @xmath740 , thus @xmath729 is a sliding minimal cone of type @xmath91 in @xmath45 with sliding boundary @xmath36 . similarly , if @xmath741 , then @xmath742 , thus @xmath729 is a sliding minimal cone of type @xmath92 in @xmath45 with sliding boundary @xmath36 . if @xmath209 , then the blow - up limit is sliding minimal cone containing @xmath36 . we know , by theorem
[ thm : mccb ] , that there are only three kinds of minimal cone which contain @xmath36 .
that is , @xmath36 or @xmath610 , here @xmath87 is a minimal cone of type @xmath91 or @xmath92 .
thus , in the case @xmath209 , there are three possible values for @xmath72 , that is @xmath743 , @xmath744 and @xmath745 . in particular , if @xmath209 and @xmath746 , then @xmath747 .
[ le : d1 ] let @xmath49 be a sliding almost minimal set , @xmath209 .
if a blow - up limit of @xmath3 at @xmath42 is the plane @xmath36 , then there exists @xmath68 such that @xmath748 . without loss of generality , we assume @xmath749 .
@xmath36 is a blow - up limit of @xmath3 at @xmath7 . by corollary 29.53 in @xcite
, we get that @xmath750 .
let @xmath85 be a small enough number , let @xmath751 be as in lemma [ mainlemma ] .
we take @xmath752 , and let @xmath753 be as in lemma [ mainlemma ] .
we take @xmath68 such that @xmath754 and @xmath755 where @xmath756 is taken as in proposition 5.24 in @xcite . by lemma [ mainlemma ] , for any @xmath757 , @xmath758 and for all ball @xmath759 , @xmath760
thus , in particular , for any @xmath761 , @xmath762 for any @xmath763 , by theorem 28.7 in @xcite , we get that @xmath764 thus @xmath765 therefore , for @xmath766 , @xmath767 by lemma [ mainlemma ] , for any @xmath768 , @xmath769 we shall deduce , from equation , that for any @xmath770 , @xmath771 once we have proved this , the desire result follows . we assume , for the sake of a contradiction , that @xmath772 let @xmath773 , and let @xmath774 be the projection of @xmath775 on @xmath36 , then @xmath776 .
we choose @xmath777 such that @xmath778 from equation , we can get that @xmath779 absurd .
[ le : removeboundary ] let @xmath45 , @xmath36 and @xmath20 be as in lemma [ mainlemma ] , let @xmath44 be a @xmath28-sliding - almost - minimal set with @xmath209 .
let @xmath780 .
then @xmath781 , and @xmath436 is also @xmath28-sliding - almost - minimal .
we put @xmath782 .
let @xmath783 .
we assume , for the sake of contradiction , that @xmath784 .
since @xmath785 is a subset of @xmath36 , it is rectifiable , thus for @xmath786-a.e .
@xmath787 , @xmath788 .
without loss of generality , we suppose that @xmath789 , then there exists a radius @xmath790 such that for all @xmath791 , @xmath792 since @xmath3 is sliding almost minimal , by theorem 28.7 ( almost monotonicity of density property ) in @xcite , we can find a radius @xmath793 such that for all @xmath794 , @xmath795 since @xmath3 is sliding almost minimal and @xmath209 , by lemma [ mainlemma ] , there exists @xmath68 such that for any @xmath796 , there exists a sliding minimal cone @xmath797 such that @xmath798 and for any ball @xmath759 , @xmath799 we take @xmath800 , and consider a collection of balls @xmath801 it is a vitali covering for @xmath802 . by a vitali s covering theorem for the hausdorff measure , see for example , there exists a finite or countably infinite disjoint subcollection @xmath803 such that @xmath804 we now consider two balls @xmath805 and @xmath806 , where @xmath807 , @xmath808 and @xmath809 .
we can see that @xmath810 and @xmath811 , thus by equation , we can get that @xmath812 and @xmath813 it is very easy to see that @xmath814 is a family of disjoint balls and @xmath815 for @xmath816 , we denote @xmath817 , then @xmath818 and @xmath819 ; otherwise , @xmath820 , any blow - up limit of @xmath3 at @xmath821 must be @xmath36 , and by lemma [ le : d1 ] , there is a small ball @xmath822 such that @xmath823 , that is impossible . by our choice of @xmath824 , we have that @xmath825 thus @xmath826 and combine with equations and , to obtain @xmath827 since @xmath828 , we have that @xmath829 , thus @xmath830 or @xmath831 . if @xmath830 , the sliding minimal @xmath832 which we chose in can be written @xmath833 , where @xmath834 is a sliding minimal cone of type @xmath91 . in this case , @xmath835 , @xmath639 , are two disks with radius @xmath836 , thus @xmath837 combine this equation with equations , , and , we can get that @xmath838 but from equation , we can get that @xmath839 which contradict with equation , because @xmath840 .
if @xmath831 , a very similar calculation as above case , we can get that @xmath841 and @xmath842 but from equation , we obtain that @xmath843 we also get a contradiction .
we proved that @xmath781 .
we will go to prove that @xmath436 is also sliding almost minimal .
let @xmath21 be any @xmath19-sliding - deformation .
since @xmath3 is @xmath28-sliding - almost - minimal , applying proposition 20.9 in @xcite , we get that @xmath844 since @xmath845 , we can get that @xmath846 we know that @xmath780 and @xmath781 , thus @xmath847 similarly , we can get that @xmath848 thus @xmath849 from inequalities , and , we obtain that @xmath850 applying proposition 20.9 in @xcite , we get that @xmath436 is @xmath28-sliding - almost - minimal .
if @xmath45 , @xmath36 , @xmath20 , @xmath3 and @xmath436 are as in lemma [ le : removeboundary ] , and we suppose that @xmath851 , then @xmath852 can only take two values @xmath853 and @xmath854 . indeed , since @xmath209 , any blow - up limit @xmath87 of @xmath3 at @xmath7 is a sliding minimal cone which contains the boundary @xmath36 , thus @xmath855 or @xmath856 , @xmath251 is a sliding minimal cone of type @xmath91 or @xmath92 , hence the density @xmath857 can only take three values , @xmath743 , @xmath744 and @xmath745 .
but if @xmath750 , then by lemma [ le : d1 ] , we can see that @xmath858
. therefore , @xmath830 or @xmath745 .
we see that @xmath859 thus @xmath860 and @xmath861 or @xmath854 .
[ le : bdap ] let @xmath45 and @xmath36 be as in , @xmath862 be the orthogonal projection onto @xmath36 .
suppose that @xmath20 is an open set , @xmath49 is @xmath28-sliding - almost - minimal , @xmath863 and @xmath864 or @xmath854 .
if @xmath187 is small enough , @xmath865 , and for any @xmath866 , and any @xmath867 , there exists a sliding minimal cone @xmath729 of type @xmath91 or @xmath92 centered at @xmath32 , such that @xmath868 then for any @xmath869 , we can find a point @xmath870 such that @xmath871 for any @xmath869 , we put @xmath872 .
we take a point @xmath873 such that @xmath874 if @xmath875 , @xmath876 , then nothing needs to be done .
if @xmath877 , we claim that @xmath878 is a point which we desire .
it is quite easy to see that @xmath879 ; otherwise @xmath880 this gives a contradiction .
we put @xmath881 .
since @xmath882 , and @xmath883 is perpendicular to @xmath36 , we can find @xmath884 such that @xmath885 .
we claim that @xmath886 ; once we have proved our claim , we can get that @xmath887 we assume , for the sake of a contradiction , that @xmath888 , then @xmath889 if @xmath890 , we take @xmath891 , then @xmath892 and @xmath893 thus @xmath894 this is a contradiction . if @xmath895 , we can construct a projection to show that @xmath3 is not almost minimal .
[ lemmaplane ] let @xmath45 , @xmath36 be as in , u an open set with @xmath896 .
suppose that @xmath44 is @xmath28-sliding - almost - minimal .
if @xmath864 , then for each small @xmath85 , we can find a radius @xmath68 , a sliding minimal cone @xmath87 of type @xmath91 and a bihlder map @xmath897 such that @xmath898 we can assume that @xmath20 is an open ball @xmath899 for some @xmath900 .
for any @xmath901 $ ] , we let @xmath751 be as in lemma [ mainlemma ] , we suppose that @xmath195 is so small , that @xmath902 where @xmath756 is taken as in proposition 5.24 in @xcite , and @xmath589 is taken as in theorem 28.7 in @xcite .
let @xmath903 and @xmath753 be as in lemma [ mainlemma ] and such that @xmath904 and @xmath905 we take @xmath906 , and let @xmath907 , @xmath908 be also as in lemma [ mainlemma ] .
we always suppose that @xmath909 . by theorem 28.7 in @xcite
, we can find @xmath910 such that @xmath911 where @xmath912 by using lemma [ mainlemma ] , for any @xmath913 $ ] , there exists a minimal cone @xmath914 of type @xmath91 center at @xmath7 such that @xmath915 and for any ball @xmath916 , @xmath917 first , we consider any point @xmath918 . by
, if we take @xmath919 , we will get that @xmath920 from this inequality , by taking @xmath921 , we can get that @xmath922 by using theorem 28.7 in @xcite , we get that @xmath923 thus @xmath924 by lemma [ mainlemma ] , we can find minimal sliding cone @xmath729 for any @xmath925 such that @xmath926 and for any ball @xmath927 , @xmath928 we now consider any point @xmath929 . from lemma [ le : bdap ]
, we get that @xmath930 thus @xmath931 we take @xmath932 , and @xmath933 , then @xmath934 .
we take @xmath729 as in , then @xmath935 , thus @xmath936 hence @xmath937 but we know that @xmath938 hence by using a monotonicity property , proposition 5.24 in @xcite , we have that @xmath939 thus @xmath940 for any @xmath941 $ ] , we can apply lemma 16.11 in @xcite , there exists a plane @xmath942 through @xmath775 such that @xmath943 for any @xmath944 $ ] , we put @xmath945 , then @xmath946 , and @xmath947 .
we take @xmath948 as in , then @xmath949 we do not know whether or not the sliding minimal cone @xmath948 passes through the point @xmath775 , but we can do a translation of @xmath948 such that it is centered at @xmath950 , we denote it by @xmath942 , i.e. @xmath951 . then @xmath942 is a sliding minimal cone contains @xmath775 , and @xmath952 it follows from and that , for any @xmath953 , for any @xmath954 $ ] , there exist a cone @xmath942 such that @xmath955 where @xmath942 is a plane when @xmath75 is small , @xmath956 is a half plane when @xmath75 is large . from the inequalities and , we get that , for any @xmath957 and any @xmath954 $ ] , we can find a cone @xmath956 though @xmath32 such that @xmath958 where @xmath956 is a minimal cone when @xmath959 , and @xmath956 is a sliding minimal cone of type @xmath91 when @xmath960 . by corollary [ co : pt ] , we can find a bihlder map @xmath961 and a sliding minimal cone @xmath962 of type @xmath91 such that holds with @xmath963 . [
le : pufp2 ] let @xmath45 , @xmath36 be as in , u an open set with @xmath896 .
suppose that @xmath44 is @xmath964-sliding - almost - minimal .
if @xmath830 , then for each small @xmath85 , we can find a radius @xmath68 , a bihlder map @xmath965 and a sliding minimal cone @xmath87 of type @xmath91 such that @xmath966 we put @xmath780 , then @xmath436 is also @xmath967-sliding - almost - minimal . by lemma [ lemmaplane ] , for each small @xmath85 , we can find @xmath68 , a bihlder map @xmath968 and a sliding minimal cone @xmath87 of type @xmath91 such that @xmath969 thus @xmath970 [ re : plane ] suppose that @xmath45 , @xmath36 and @xmath20 are as in lemma [ lemmaplane ] , and that @xmath44 is a @xmath28-sliding - almost - minimal set with @xmath864 or @xmath744 . if @xmath971 is small enough , we can find @xmath972 such that when the radius @xmath68 is such that @xmath973 then for any @xmath974 and any @xmath975 , we can find a cone or sliding minimal cone @xmath976 that depends on @xmath977 such that @xmath978 where @xmath976 is a minimal cone when @xmath979 , and @xmath976 is a sliding minimal cone when @xmath980 . indeed , when we look at the proof of lemma [ lemmaplane ] , we let @xmath971 be such that @xmath981 then we take @xmath982 and let @xmath983 be as in lemma [ mainlemma ] . finally , @xmath984 will be what we desire .
[ lemmay ] let @xmath45 , @xmath36 be as in , @xmath20 an open set with @xmath896 .
suppose that @xmath985 is an @xmath28-sliding - almost - minimal set .
if @xmath986 , then for each small @xmath85 , we can find a radius @xmath68 , a bihlder map @xmath987 and a sliding minimal cone of type @xmath92 such that @xmath988 as in lemma [ lemmaplane ] , we can assume @xmath20 is an open ball @xmath899 for some @xmath900 . let @xmath85 be a positive number , and let @xmath751 be as in lemma [ mainlemma ] ; we suppose @xmath195 small enough so that @xmath989 where @xmath756 is taken as in proposition 5.24 in @xcite , and @xmath589 is taken as in theorem 28.7 in @xcite .
let @xmath903 and @xmath753 be as in lemma [ mainlemma ] so that @xmath904 and @xmath990 where @xmath991 is the constant which is considered in lemma 14.12 in @xcite .
we take @xmath992 .
let @xmath993 and @xmath983 be as in lemma [ mainlemma ] .
we suppose that @xmath909 . as in the proof of lemma [ mainlemma ]
, we put @xmath780 , then @xmath436 is also @xmath967-sliding - almost - minimal . by theorem 28.7 in @xcite
, there exist @xmath994 such that @xmath995 and @xmath996 where @xmath912 by using lemma [ mainlemma ] , for any @xmath997 $ ] there exists a minimal cone @xmath834 of type @xmath92 center at @xmath7 such that @xmath998 first , for any @xmath999 , we take @xmath1000 and @xmath1001 , then by , we have @xmath1002 from this inequality , we can get that @xmath1003 applying theorem 28.7 in @xcite , we get that @xmath1004 thus @xmath1005 taking @xmath1006 , we get from that @xmath1007 by lemma [ mainlemma ] , for any @xmath1008 , there exists a sliding minimal cone @xmath1009 centered at @xmath32 of type @xmath91 such that @xmath1010 and for any ball @xmath1011 , @xmath1012 for @xmath1013 , we have that @xmath1014 since @xmath1015 , there exists a point @xmath1016 such that @xmath1017 .
we take @xmath1018 , that is a translation of @xmath1019 ; it is a siliding minimal through the point @xmath32 , and @xmath1020 it follows from and that , for any @xmath1021 , and any @xmath1022 , there exists a sliding minimal cone @xmath1023 centered at @xmath32 , either of type @xmath91 or of type @xmath92 , such that @xmath1024 next , we consider @xmath1025 .
if @xmath1026 , we take a point @xmath1027 such that @xmath1028 we take
@xmath1029 and @xmath1030 , then @xmath1031 thus @xmath1032 and @xmath1033 since @xmath1034 , we have that @xmath1035 thus @xmath1036 hence @xmath1037 consider the sliding minimal cone @xmath883 as in ; it is a minimal cone centered at point @xmath878 of type @xmath91 . since @xmath1038
, we deduce from that @xmath1039 thus @xmath1040 applying proposition 5.24 in @xcite , we can get that @xmath1041 thus @xmath1042 by lemma 16.11 in @xcite , for any @xmath1043 $ ] , there exist a plane @xmath1044 through @xmath775 such that @xmath1045 for @xmath1046 $ ] , we put @xmath1047 , then @xmath1048 consider the sliding minimal cone @xmath1049 as in , we can get that @xmath1050 we now take @xmath1051 it is a sliding minimal cone centered at @xmath950 , thus through the point @xmath775 , which satisfy that @xmath1052 it follows from and that , in the case @xmath1053 and @xmath1054 , for @xmath1055 , we can find a cone @xmath1044 such that @xmath1056 where @xmath1044 is a minimal cone when @xmath1057 is small , and @xmath1044 is a sliding minimal cone when @xmath1057 is large .
we now consider the case when @xmath1025 with @xmath1058 .
we take @xmath1059 , and put @xmath1060 , then @xmath1061 .
we take @xmath1062 a minimal cone as in , then we can get that @xmath1063 thus @xmath1064 applying proposition 5.24 in @xcite , we get that @xmath1065 thus @xmath1066 or @xmath1067 .
if @xmath1068 , then for any @xmath1069 , by using lemma 16.11 in @xcite , there exists a minimal cone @xmath1044 centered at @xmath775 of type @xmath33 such that @xmath1070 and for any ball @xmath927 @xmath1071 for any @xmath1072 $ ] , we put @xmath1073 , then @xmath1074 , and @xmath1075 , thus @xmath1076 .
let @xmath1077 be the sliding minimal cone which is considered in , then we can get that @xmath1078 we take a point @xmath1079 such that @xmath1080 , and take @xmath1081 , which through point @xmath775 , we obtain that @xmath1082 it follows from and that , when @xmath1053 and @xmath1083 with @xmath1067 , for any @xmath1084 $ ] , we can find a cone @xmath1044 such that @xmath1085 where @xmath1044 is a minimal cone when @xmath1057 small , and @xmath1044 is a sliding minimal cone with boundary @xmath36 when @xmath1057 large .
if @xmath1066 .
we set @xmath1086 and denote @xmath1087 . using the same argument as in lemma 16.25 in (
* in page 205 ) , we get that for @xmath1088 $ ] , there is a plane @xmath1044 through @xmath32 such that @xmath1089 for @xmath1090 $ ] , we take a point @xmath1091 such that @xmath1092 and consider the cone @xmath1093 as in , where @xmath1094
. we can get that @xmath1095 by a similar argument as before , we can find a cone @xmath1044 which is a translation of @xmath1093 such that @xmath1096 it follows that , when @xmath1053 and @xmath1083 with @xmath1066 , for any @xmath1097 $ ] , we can find a cone @xmath1044 such that @xmath1098 where @xmath1044 is a minimal cone when @xmath1057 is small , and @xmath1044 is a sliding minimal cone when @xmath1057 is large . from inequalyties , , and , we can say that for any @xmath1099 , and for any @xmath1097 $ ] , there is a cone @xmath1044 such that @xmath1100 where @xmath1044 is a minimal cone when @xmath1101 , and @xmath1044 is a sliding minimal cone when @xmath1102 . by corollary [ co : pt ]
, we can get our desired result .
[ lemmay2 ] let @xmath45 , @xmath36 be as in , @xmath20 an open set with @xmath896 .
suppose that @xmath44 is an @xmath28-sliding - almost - minimal set . if @xmath831 , then for each small @xmath85 , we can find a radius @xmath68 , a bihlder map @xmath968 and a sliding minimal cone @xmath87 of type @xmath92 such that @xmath1103 we put @xmath780 , then @xmath436 is also @xmath967-sliding - almost - minimal . by lemma [ lemmay ] , for each small @xmath85 , we can find @xmath68 , a bihlder map @xmath968 and a sliding minimal cone @xmath87 of type @xmath92 such that @xmath1104 thus @xmath970 [ re : y ] suppose that @xmath45 , @xmath36 and @xmath20 are as in lemma [ lemmay ] , and that @xmath44 is a @xmath28-sliding - almost - minimal set with @xmath1105 or @xmath745 . if @xmath971 is small enough , we can find @xmath972 such that when @xmath68 is such that @xmath1106 then for any @xmath974 and any @xmath975 , we can find a minimal cone or sliding minimal cone @xmath976 such that @xmath978 where @xmath976 is a minimal cone when @xmath979 , and @xmath976 is a sliding minimal cone when @xmath980 . indeed , we can take @xmath971 be such that @xmath1107 then we take @xmath982 and let @xmath983 be as in lemma [ mainlemma ] .
we can check from the proof of lemma [ lemmay ] that @xmath984 is the number what we desire .
[ prop : prb ] let @xmath45 , @xmath36 be as in , @xmath20 an open set .
let @xmath49 be an @xmath28-sliding - almost - minimal set , and @xmath1108 be a point .
if @xmath1109 , then for each small @xmath85 , we can find a radius @xmath68 , a sliding minimal cone @xmath87 centered at @xmath32 and a bihlder map @xmath1110 such that @xmath1111 in addition , if @xmath739 , @xmath87 is a cone of type @xmath91 ; if @xmath741 , @xmath87 is a cone of type @xmath92 ; if @xmath1112 , @xmath250 where @xmath251 is a cone of type @xmath91 ; if @xmath1113 , @xmath1114 where @xmath251 is a cone of type @xmath92 .
the proof immediately follows from lemma [ lemmaplane ] , lemma [ le : pufp2 ] , lemma [ lemmay ] and lemma [ lemmay2 ] .
[ co : regularity ] let @xmath45 , @xmath36 be as in , @xmath20 an open set .
let @xmath49 be an @xmath28-sliding - almost - minimal set with @xmath177 .
then for each small @xmath85 and each @xmath1115 , we can find a radius @xmath68 , a sliding minimal cone @xmath87 and a bihlder map @xmath1116 such that @xmath1117 since @xmath177 , any blow - up limit @xmath436 of @xmath3 at @xmath32 contains @xmath36 , so it is a sliding minimal cone contains @xmath36 . by theorem [ thm : mccb ] , we can get that @xmath1118 or @xmath1119 , where @xmath87 is a cone of type @xmath91 or @xmath92 . if @xmath1118 , by lemma [ le : d1 ]
, then there exists a ball @xmath286 such that @xmath748 , thus hold automatically . if @xmath1120 , then @xmath1119 where @xmath87 is a sliding minimal cone of type @xmath91 or @xmath92 ; we get that @xmath1112 or @xmath745 , by proposition [ prop : prb ] , we obtain the conclusion . in the previous section
, we get some regularity for sliding almost minimal sets with whose boundary is a plane . in this section
we will give a similar result , but with where the boundary is a @xmath12 manifold .
let @xmath81 be a connected closed set such that the boundary @xmath82 is a @xmath11-dimensional @xmath12 manifold .
for any @xmath1121 , the tangent cone of @xmath1122 at @xmath32 is a half space , and the boundary of the half space is the tangent plane of @xmath82 at @xmath32 .
[ mainthm ] let @xmath1122 be as above , @xmath1123 , @xmath20 be a neighborhood of @xmath32 .
suppose that @xmath83 is an @xmath28-sliding - almost - minimal set with sliding boundary @xmath82 and that @xmath84 .
then for each small @xmath85 , we can find a radius @xmath86 , a sliding minimal cone @xmath87 in @xmath45 with sliding boundary @xmath36 and a bihlder map @xmath88 such that @xmath1124 where @xmath45 is the tangent cone of @xmath1122 at @xmath32 and @xmath36 is the boundary of @xmath45 .
the strategy of the proof will be the same as for corollary [ co : regularity ] .
we do not want repeat the whole section above , because most of the statements and proofs still work .
we only give a sketch .
firstly , lemma [ mainlemma ] is still true when we replace @xmath45 and @xmath36 by @xmath1122 and @xmath82 respectively .
that is , it can be stated as follows : [ mainlemma2 ] let @xmath1122 and @xmath82 be as in theorem [ mainthm ] .
suppose that @xmath1125 is @xmath28-sliding - almost - minimal . if for each @xmath85 , we can find @xmath1126 such that if @xmath1127 and @xmath68 are such that @xmath1128 then for every @xmath1129 $ ] there is a sliding minimal cone @xmath1130 such that @xmath1131 and for any ball @xmath927 , @xmath1132 where @xmath729 is a sliding minimal cone in @xmath1133 with sliding boundary @xmath1134 , where we denote by @xmath1133 and @xmath1134 the tangent cone of @xmath1122 at @xmath32 and tangent plane of @xmath82 at @xmath32 respectively .
if @xmath84 , then we can suppose that @xmath1135 . for each @xmath1136 , we see that any blow - up limit @xmath87 of @xmath3 at @xmath32 is a sliding minimal cone in @xmath1133 with sliding boundary @xmath1134 , see ( * ? ? ?
* theorem 24.13 ) .
if @xmath84 , we have that @xmath1137 , thus @xmath1138 or @xmath1139 , where @xmath251 is a sliding minimal cone in @xmath1133 with sliding boundary @xmath1134 of type @xmath91 or @xmath92 .
hence , we get that @xmath746 , @xmath744 or @xmath745 . similar to lemma [ le : d1 ] , we can get that if @xmath93 is sliding almost minimal and a blow - up limit of @xmath3 at @xmath1123 is the tangent plane @xmath1134 of @xmath82 at that point , then there exists @xmath68 such that @xmath1140 .
once we get that , we can show a similar result to lemma [ le : removeboundary ] .
that is , if @xmath1125 is @xmath28-sliding - almost - minimal and @xmath93 , then , by putting @xmath1141 , we shall have @xmath1142 and @xmath436 is also @xmath28-sliding - almost - minimal .
thus , @xmath1143 can only take two values .
that is , @xmath853 and @xmath854 .
finally , if @xmath1125 is sliding almost minimal , @xmath1127 , we can get that if @xmath739 or @xmath854 , then the sliding minimal cone @xmath729 which is taken in lemma [ mainlemma2 ] is of type @xmath91 or @xmath92 ; if @xmath94 and @xmath746 , then @xmath1144 ; if @xmath1112 or @xmath745 , then @xmath1145 , @xmath87 is of type @xmath91 or @xmath92 .
we also need a lemma like lemma [ le : bdap ] .
let @xmath1122 and @xmath82 be as in lemma [ mainlemma2 ] , @xmath20 be an open set .
let @xmath1125 be a @xmath28-sliding - almost - minimal set .
let @xmath1146 be a small number .
suppose that @xmath1136 , @xmath739 or @xmath854 .
if @xmath1147 , and for any @xmath1148 and any @xmath1149 , there exists a sliding minimal cone @xmath1150 in @xmath1151 ( the tangent cone of @xmath1122 at @xmath774 ) with sliding boundary @xmath1152 such that @xmath1153 then there exists a radius @xmath68 such that for any @xmath1154 , we can find a point @xmath1155 satisfying @xmath1156 now , we state a similar result as lemma [ lemmaplane ] and lemma [ lemmay ] , or rather , a similar result as remark [ re : plane ] and remark [ re : y ] .
the proof can be adapted from the proof of lemma [ lemmaplane ] and lemma [ lemmay ] .
[ le : appbc ] let @xmath1122 and @xmath82 be as in lemma [ mainlemma2 ] .
let @xmath1125 be a @xmath28-sliding - almost - minimal set such that @xmath1157 , @xmath1158 .
if @xmath971 is a small enough number , then we can find @xmath1159 such that when @xmath1160 for some @xmath68 , we have that for any @xmath1161 , and any @xmath1162 , there exists a cone or a sliding minimal cone @xmath718 satisfying @xmath1163 where @xmath718 is a cone when @xmath1164 , @xmath718 is a sliding minimal cone centered at a point in @xmath1165 when @xmath1166 . without loss of generality , we assume @xmath749 .
let @xmath45 be the tangent cone of @xmath1122 at @xmath7 , @xmath36 be the tangent plane of @xmath82 at @xmath7 .
then @xmath45 is a half space , and @xmath36 is its boundary .
let @xmath85 and @xmath68 be as in lemma [ le : appbc ] . since @xmath82 is a 2-dimensional @xmath12 manifold , for any @xmath1167 , we can find a radius @xmath1168 and a @xmath12 diffeomorphism @xmath1169 such that @xmath1170 , @xmath1171 , @xmath1172 and @xmath1173 we put @xmath1174 for any @xmath1175 and @xmath975 , by lemma [ le : appbc ] , we can find a minimal cone or a sliding minimal cone @xmath1176 such that @xmath1177 then @xmath1178 where we assume that @xmath1176 is centered at @xmath878 , and denote @xmath1179 we note from lemma [ le : appbc ] that if @xmath1176 is a sliding minimal cone , then it is centered at a point in @xmath82 .
thus @xmath1180 , and @xmath1176 is a sliding minimal cone in @xmath1181 with sliding boundary @xmath1182 . since @xmath1183
, we have @xmath1184 .
we take @xmath1185 , then @xmath1186 thus @xmath1187 @xmath1188 is a minimal cone or a sliding minimal .
let @xmath1189 be the translation which send point @xmath775 to @xmath1190 . then @xmath1191 .
if @xmath1176 is a sliding minimal cone , then @xmath1188 is a slding minimal cone in @xmath1192 with sliding boundary @xmath1193 .
we put @xmath1194 , then it is quite easy to see that @xmath1195 maps @xmath45 and @xmath36 to @xmath1192 and @xmath1193 respectively .
since @xmath1196 , we can find a rotation @xmath1197 centered at point @xmath774 , which will rotate @xmath1192 and @xmath1193 to @xmath45 and @xmath36 respectively , such that @xmath1198 then @xmath1199 is a sliding minimal cone in @xmath45 with sliding boundary @xmath36 .
we take @xmath1200 when @xmath1188 is a slding minimal cone , and take @xmath1201 when @xmath1188 is a minimal cone , then @xmath1202 by corollary [ co : pt ] , we can find a radius @xmath1203 , a sliding minimal cone @xmath87 in @xmath45 with sliding boundary @xmath36 , and a bihlder map @xmath1204 such that @xmath1205 we now take @xmath1206 . then @xmath1207 is a bihlder map , and we can easily check that hold if we take @xmath1208 .
let @xmath81 be a connected closed set such that the boundary @xmath82 is a @xmath11-dimensional connected compact @xmath12 manifold .
let @xmath785 be any abelian group , @xmath1209 be a subgroup of the ech homology group @xmath1210 .
we say a compact set @xmath84 spans @xmath1209 if @xmath1209 is contained in the kernel of the homomorphism induced by the inclusion map @xmath1211 .
we set @xmath1212 from paper @xcite , we see that there exist a set @xmath1213 , we call it a ech minimizer , such that @xmath1214 let s check that @xmath5 is also sliding minimal with boundary @xmath82 . let @xmath21 be any sliding - deformation in @xmath1122 .
we put @xmath1215 , denote by @xmath1216 and @xmath1217 the inclusion maps .
we consider the map @xmath1218\to f , \psi(x , t)=\varphi_t(x).\ ] ] it is continuous , and @xmath1219 , @xmath1220 , thus the maps @xmath1217 and @xmath1221 are homotopy equivalent
. then @xmath1222 , where for any map between two topology spaces @xmath1223 , we denote by @xmath1224 the homomorphism @xmath1225 induced by the map @xmath1226 . however , we know that @xmath1227 , thus @xmath1228 but we know that @xmath1229 , thus @xmath1230 , and @xmath1231 , so @xmath1232 @xmath5 is sliding minimal .
we now consider an analogous topic , that replace ech homology by singular homology . since @xmath82 is a two dimensional @xmath12 manifold ,
the singular homology groups and ech homology groups coincide , that is , @xmath1233 .
we say that a compact subsets @xmath93 spans @xmath1209 in singular homology , if @xmath1209 is contained in the kernel of the homomorphism @xmath1234 induced by the inclusion map @xmath1211 .
we consider another collection of compact sets @xmath1235 it is quite easy to see that @xmath1236 , that is because there is a canonical homomorphism from singular homology group to ech homology group @xmath1237 , and the following diagram commutes : @xmath1238\ar@{=}[d]&h_1(e;g)\ar[d]\\ \check{h}_1(\partial\sigma;g)\ar[r]&\check{h}_1(e;g ) . }
\ ] ] if @xmath3 spans @xmath1209 in singular homology , then from the above commutative diagram , we can get that @xmath3 spans @xmath1209 in ech homology , thus @xmath1239 .
our goal is to find a singular minimizer , that is , we want to find a set @xmath1240 , we call it a singular minimizer , such that @xmath1241 let @xmath5 be a ech minimizer . we know , from above discussion , that @xmath5 is sliding minimal . thus for any @xmath1242 , there is a neighborhood of @xmath32 where @xmath5 is bihlder equivalent to minimal cone and by a bihlder mapping that preserves @xmath82 . by a same argument as in (
* section 6 ) , we conclude that @xmath5 is hlder neighborhood retract
. let s check that @xmath5 is a singular minimizer , it is sufficient to show that @xmath5 spans @xmath1209 in singular homology .
indeed , the canonical homomorphism @xmath1243 is an isomorphism since @xmath5 is neighborhood retract , see for example @xcite .
now @xmath5 is a ech minimizer , @xmath5 spans @xmath1209 in ech homology , thus @xmath5 spans @xmath1209 in singular homology , and we get the conclusion . | in @xcite , jean taylor has proved a regularity theorem away from boundary for almgren almost minimal sets of dimensional two in @xmath0 .
it is quite important for understanding the soap films and the solutions of plateau s problem away from boundary . in this paper
, we will give a regularity result on the boundary for two dimensional sliding almost minimal sets in @xmath0 .
it will be of use for understanding their boundary behavior . |
we consider an investment model in continuous time where the decision maker has the option to invest in a given project yielding uncertain returns @xmath0 .
the investors objective is to choose the entry time such that a particular objective functional ( often of either discounted or ergodic type ) is maximised . at the time of the entry , a known fixed lump sum @xmath1 of entry costs must be paid .
once the entry is made , the investment incurs a known constant instantaneous running cost @xmath2 . in the classical perpetual version of this problem , see , e.g. , @xcite , it is assumed that once the entry is made , the return from the investment will accrue from the investment date to infinity .
a variant of this problem includes the possibility of voluntary exits , see , e.g. , @xcite .
the problem becomes then of sequential nature , where a sequence of entry and exit times is determined such that the objective functional is maximised .
the purpose of this paper is to study a class of multiple entry problems where the exits are not voluntary but forced .
this type of problem was first studied in @xcite and can be informally described as follows .
subject to return uncertainty modelled by a time homogeneous diffusion process @xmath0 , and known entry and running costs @xmath1 and @xmath2 , the investor chooses the time of entry . however , the investment is subject to exogenous interventions , which will terminate the flow of returns .
these interventions occur uniformly over time and are modelled by the jumps of a poisson process @xmath3 independent of @xmath0 .
once the investment is terminated , the decision maker can make a new entry .
the objective is then to maximise the expected present value of the total revenue from the investment .
as was discussed already in @xcite , this setting lends itself to a possible application to a so - called liquidation risk @xcite . consider the case where the investor is funding the investment with borrowed money . to increase liquidity on the lenders side , it is possible that the lender is given ( or requires ) the right to seize the asset and put it to alternative use , that is , liquidate .
thus the intervention would be forced liquidation from the lenders side .
after a forced liquidation , the investor can find a new lender to make a new entry .
this paper makes two contributions .
first , we allow the underlying stochastic process @xmath0 to follow a general one - dimensional diffusion process with natural boundaries . in comparison to @xcite , where the case of geometric brownian motion is considered ,
this is a substantial generalisation which has not , to our best knowledge , been studied earlier in the literature .
furthermore , we introduce a so - called catastrophe risk in the model as follows : we attach iid bernoulli trials to the jump times of @xmath3 and if the trial fails , no further re - entries are allowed . in the liquidation risk application described above , the catastrophe event describes a fundamental change in the economical environment of the investment opportunity such that all lenders lose interest in financing a new entry .
such a change could be , for instance , due to the emergence of a new technology .
this is again a substantial generalisation and it effectively means that the number of forced exits up to the catastrophe is geometrically distributed . somewhat remarkably we find that the optimal investment threshold is independent of the success probability of the bernoulli trials . the paper is organised as follows .
in section 2 we set up the probabilistic framework . in section 3
the entry problem is introduced .
the solution is derived in section 4 .
the paper is concluded in section 5 with illustrative examples .
we set up the probabilistic framework for the entry problem , for details , see @xcite .
let @xmath4 , where @xmath5 , be a complete filtered probability space satisfying the usual conditions .
assume that the filtration is rich enough to carry the underlying state process @xmath6 and a poisson process @xmath7 .
we assume that the process @xmath3 has jump intensity @xmath8 and that it is independent of @xmath0 .
the process @xmath0 is a linear diffusion evolving on @xmath9 with the infinitesimal generator @xmath10 and initial state @xmath11 . in what follows , we assume that the functions @xmath12 and @xmath13 are continuous and that the process does not die inside the state space .
the densities of the speed measure @xmath14 and the scale function @xmath15 of @xmath0 are defined , respectively , as @xmath16 and @xmath17 for all @xmath18 , where @xmath19 .
we denote as , respectively , @xmath20 and @xmath21 the increasing and decreasing solution of the ode @xmath22 , where @xmath23 , defined on the domain of the characteristic operator of @xmath0 . by posing appropriate boundary conditions depending on the boundary classification of @xmath0 , the functions @xmath24 and @xmath25 are defined uniquely up to a multiplicative constant and can be identified as the minimal @xmath26-excessive functions . to impose the boundary conditions , we assume , using the terminology of @xcite , that the boundaries @xmath27 and @xmath28 are natural .
this means that @xmath29 finally , we denote as @xmath30 the probability measure @xmath31 conditioned on the initial state @xmath11 and as @xmath32 the expectation with respect to @xmath30 . for @xmath23
, we denote as @xmath33 the class of real valued functions on @xmath9 satisfying the condition @xmath34<\infty$ ] . for a function @xmath35 ,
the resolvent @xmath36 is defined as @xmath37.\ ] ] it is well known , see , e.g. , @xcite , that for the considered class of diffusion processes , the resolvent operator can be expressed as @xmath38 where @xmath39 is the wronskian determinant .
we also point out that the resolvent operator satisfies the following resolvent equation @xmath40 for all @xmath41 .
[ resolvent remark ] let @xmath42 . for an arbitrary @xmath43 ,
define the functions @xmath44 and @xmath45 as @xmath46 for all @xmath47 .
then , obviously , @xmath48 and @xmath49 for all @xmath50 .
furthermore , @xmath51 and @xmath52 the following auxiliary result can be readily verified using the representation and the conditions .
[ lemma001 ] let @xmath23 and @xmath53
. then @xmath54
we are now in position to state our entry problem .
let @xmath55 denote the jump times of the poisson process @xmath3 with the convention @xmath56 .
denote as @xmath57 an arbitrary sequence of stopping times @xmath58 taking values in @xmath59 $ ] satisfying the constraint @xmath60 for all @xmath61 , where @xmath62 .
let @xmath63 be a measurable function , @xmath2 and @xmath1 two non - negative constants , and @xmath23 the constant rate of discounting .
motivated by the discussion in the introduction , consider the following multiple entry problems : @xmath64\end{aligned}\ ] ] and @xmath65 .
\end{aligned}\ ] ] where @xmath66 and independent of @xmath0 . here
, the random variable @xmath67 is realized as follows : we attach an independent bernoulli trial with success probability @xmath68 to each jump time of @xmath3 .
when the first failure occurs , the re - entry possibility expires permanently .
this problem was studied in @xcite in the case where the diffusion @xmath0 is a geometric brownian motion and no catastrophes occur , i.e. , @xmath69 .
we study the problems and under the following assumptions .
[ sa ] let the function @xmath63 be @xmath70 , continuous , non - decreasing , non - constant and satisfy @xmath71 . furthermore , let at least one of the constants @xmath2 and @xmath1 be strictly positive .
it is helpful to write the value functions and as infinite sums instead of sums of random length . to this end
, we use the well - known thinning procedure of poisson processes , see , e.g. , @xcite . as we have labeled the jump times of @xmath3 with outcomes of independent bernoulli trials , we can split the process @xmath3 into two independent poisson processes @xmath72 and @xmath73 corresponding to outcomes @xmath27 and @xmath74 , respectively .
moreover , the intensities of @xmath72 and @xmath73 are @xmath75 and @xmath76 , respectively .
now , denote the jump times of @xmath73 as @xmath77 .
these jump times correspond to events where further entries are still possible .
on the other hand , the first jump of @xmath72 will terminate the whole investment opportunity .
denote this jump time as @xmath78 denote as @xmath79 an arbitrary sequence of stopping times @xmath58 taking values in @xmath59 $ ] satisfying the constraint @xmath80 for all @xmath61 , where @xmath81 .
then we have the following lemma .
[ mainlemma ] the value functions and can be expressed as @xmath82\end{aligned}\ ] ] and @xmath83 \nonumber .
\end{aligned}\ ] ] since the variable @xmath84 and it is independent of @xmath0 , we can use the additive functional @xmath85 to impose an additional killing rate of @xmath86 on the process @xmath0 , for details , see , e.g. , @xcite .
this results in the total killing rate of @xmath87 .
the previous lemma shows that the jumps of the process @xmath73 are the forced exit times for which further re - entry opportunities remain . to simplify the notation , we denote from now on the jump times of @xmath73 simply as @xmath88 for which @xmath89 and the stopping times @xmath90 as @xmath91 for all @xmath92 .
we point out following @xcite that the optimization problems and are well defined . for every sequence @xmath57 of stopping times , we have @xmath93 since @xmath94 , we find that @xmath95 we observe from lemma [ mainlemma ] that as the parameter @xmath68 approaches zero , the total killing rate @xmath87 increases which , in turn , decreases the value .
thus we can say that generally under our assumptions , increased catastrophe risk ( as measured by decreasing @xmath68 ) decreases the values @xmath96 and @xmath97 .
we derive the candidate solutions @xmath96 and @xmath97 in a partly heuristic way .
we start by making the working assumption that the continuation region for both active and idle problem is the interval @xmath98 for some threshold @xmath99 .
consider first the idle problem and assume that @xmath100 .
then it is optimal to wait . on an infinitesimal time period @xmath101
, the process @xmath3 will jump with probability @xmath102 .
given there is a jump , success in the bernoulli experiment implies that further entry possibilities remain . by the memoryless property of geometric distribution ,
the value of the remaining entry possibilities is @xmath103 . on the other hand ,
if the bernoulli experiment fails , there is no further entry possibilities and the remaining value is zero . based on these observations , we expect that the function @xmath96 satisfies the following partly heuristic computation @xmath104 \nonumber \\ & = \lambda pg_i(x)dt + ( 1-\lambda dt)(g_i(x ) + ( \mathcal{a}-r)g_i(x)dt ) \\ & = g_i(x ) + \left[(\mathcal{a}-r)g_i(x ) + \lambda pg_i(x ) - \lambda g_i(x ) \right]dt , \nonumber\end{aligned}\ ] ] here , we have neglected the @xmath105-term .
thus we expect that the candidate @xmath96 satisfies @xmath106 for all @xmath107 . when @xmath108 , the it is optimal to make an entry , that is , to pay the cost @xmath1 and become active .
therefore , we expect that @xmath109 for all @xmath110 .
consider now the active problem . now
the decision maker can do nothing before the first forced exit .
let @xmath111 . by reasoning similarly as in
, we expect that the function @xmath97 satisfies the following , again , partly heuristic computation @xmath112 \\ & = g_a(x ) + \left [ h(x)-c + ( \mathcal{a}-r)g_a(x ) + \lambda(pg_i(x)-g_a(x ) ) \right]dt.\end{aligned}\ ] ] therefore , we expect that the candidate @xmath97 satisfies @xmath113 for all @xmath111 .
summarizing , the task is now to solve the following coupled free boundary problem : find continuously differentiable functions @xmath96 and @xmath97 and threshold @xmath114 such that @xmath115 by the definitions and , we also expect that the functions @xmath96 and @xmath97 satisfy the growth condition @xmath116 for all @xmath47 . in @xcite it is required that @xmath117 .
this additional smoothness requirement is not necessary for the results to hold as we will see in the next section . since we are looking for a function @xmath96
that is bounded in the origin , we find from the first equation of that @xmath118 for all @xmath119 , where @xmath120 is a constant .
using this , we can rewrite the third equation of as @xmath121 for all @xmath119 .
solutions to this ode can be expressed as @xmath122 again , we are looking for a solution that is bounded in the origin , so we find that @xmath123 .
since the boundary @xmath28 is natural for the diffusion @xmath0 , we can use ( * ? ? ?
* lemma 2.1 ) to rewrite the solution as @xmath124 for all @xmath119 .
next , we study the solutions on the interval @xmath125 . using the two last equations in , we find that @xmath126 for all @xmath108 .
the solutions to this ode can be written as @xmath127 the growth condition implies that @xmath128 .
furthermore , we observe that @xmath129 where @xmath130 for all @xmath47 . summarising , we have the following candidate solutions @xmath131 to determine the unknown constants @xmath120 , @xmath132 , @xmath133 , and the threshold @xmath99 , we use value - matching and smooth - pasting conditions .
first , since the candidate value @xmath96 and its derivative are continuous over @xmath99 , we find that the conditions @xmath134 must hold .
this implies that @xmath135 and , consequently , by using lemma [ lemma001 ] that @xmath136 now we can solve the constant @xmath132 by substitution .
indeed , after an application of the resolvent equation , a simplification yields @xmath137 to characterise the threshold @xmath99 , we use the smoothness of @xmath97 over @xmath99 .
first , an application of the resolvent equation yields @xmath138 using the expression for @xmath132 , we rewrite this condition as @xmath139 we find using ( * ? ? ?
* lemma 2.1 ) and lemma [ lemma001 ] that @xmath140 we can handle the other term on the right hand side of similarly . by applying lemma [ lemma001 ] also to the left hand side , we can express the condition as @xmath141 next , apply the representation and change the order of integration ; this yields the expression @xmath142 using the properties of functions @xmath143 and @xmath144 , it is elementary to verify that @xmath145 finally , using these identities , a round of simplification shows that the condition can be rewritten as @xmath146 by our assumptions , it is obvious that there is a unique @xmath99 satisfying this condition .
summarising , we collect the results derived in this section to the following lemma .
[ candilemma ] let assumptions [ sa ] hold
. then the free boundary problem has a unique solution @xmath147 , where the functions @xmath96 and @xmath97 are defined in such that and hold and the threshold @xmath99 is characterised by .
we notice from the condition that the optimal entry threshold does not depend on the parameter @xmath68 .
this is a somewhat remarkable result . indeed , if the decision maker is idle , her decision rule is independent of the success probability of the bernoulli trial at the next jump time of the poisson process @xmath3 . even if this probability is very low , meaning that it is very likely that there will be no further re - entry opportunities after the next jump of @xmath3 , it is still optimal for the decision maker to act as if this probability is equal to one .
the purpose of this section is to prove that our candidate solution described in lemma [ candilemma ] is the solution of the main problem .
the next result is our main theorem .
[ mainthrm ] let assumption [ sa ] hold . * if @xmath148 , then @xmath149 and , and it is optimal never to make an entry .
* let @xmath150 and @xmath147 be the solution given in lemma [ candilemma ] .
then @xmath151 for all @xmath152 .
furthermore , the sequence of optimal entry times is recursively defined as @xmath153 for every @xmath154 , where @xmath155 and @xmath156 for @xmath154 . before proving the main theorem
, we establish some auxiliary results needed in the proof .
the first one shows essentially that the candidates @xmath96 and @xmath97 satisfy bellman s principle .
[ bellmanlemma ] let @xmath157 be a stopping time and @xmath158 where the inter - arrival times @xmath159
. then @xmath160.\ ] ] strong markov property of @xmath0 coupled with the memoryless property of the exponential distribution guarantees that it is sufficient to show that @xmath161,\end{aligned}\ ] ] where @xmath162 , for all @xmath152 . using the resolvent equation ,
this can be rewritten as @xmath163 where @xmath164 .
first , let @xmath165 .
using remark [ resolvent remark ] , we find that @xmath166 by substituting the expression for @xmath96 from , we obtain by using first ( * ? ? ?
* lemma 2.1 ) , then the optimality condition and finally ( * ? ? ?
* lemma 2.1 ) again @xmath167 now , let @xmath168 .
again , by using remark [ resolvent remark ] , we find that @xmath169 by substituting the expression for @xmath96 from , we find by using first the optimality condition , then ( * ? ? ?
* lemma 2.1 ) , and finally the resolvent equation , that @xmath170 this completes the proof . in @xcite
, this results was proved using the free boundary problem as a variational inequality . there , the fact that the candidates solve the variational inequality was primary and the form of the actual solution was secondary . in , we did the opposite and proved the result by the properties of the solution itself .
[ suffiremark ] since the function @xmath96 is finite and @xmath87-excessive , the process @xmath171 is a non - negative supermartingale , see , e.g. , @xcite .
furthermore , let @xmath157 be a stopping time and @xmath172 we claim that @xmath173.\ ] ] by strong markov property , we can take @xmath174 . if @xmath175 , the claim holds trivially
let @xmath100 . since @xmath176 is @xmath177-harmonic and @xmath0 has continuous paths , we have that @xmath178 = \mathbf{e}_{x}\left [ e^{-(r+(1-p)\lambda){\tau_{x^*}}}\right ] c_{i,1}\psi_{r+(1-p)\lambda}(x^ * ) = g_i(x);\ ] ] for the last equality , see , e.g. , @xcite .
[ limitlemma ] let @xmath58 be a sequence of stopping times such that @xmath179 as @xmath180 .
then @xmath181 \rightarrow 0 $ ] as @xmath180 . for each @xmath92
, we write @xmath182 = \psi_{r+(1-p)\lambda}(x)\hat{\mathbf{e}}_x \left [ \frac{g_i(x_{\tau_{j}})}{\psi_{r+(1-p)\lambda}(x_{\tau_{j}})}\right],\end{aligned}\ ] ] where @xmath183 is the expectation with respect to the probability associated with doob s @xmath176-transform of @xmath0 , see , e.g. , @xcite .
we find using the representation that @xmath184 as @xmath185 .
since the @xmath176-transform of @xmath0 is the process @xmath0 conditioned to exit the state space via the upper boundary @xmath28 , the claim follows .
we prove first case ( 2 ) .
let @xmath186 be an arbitrary sequence of stopping times such that @xmath187 where @xmath188 and the inter - arrival times @xmath159 .
by using the supermartingale property of @xmath171 , the fact that @xmath189 for all @xmath47 , then lemma [ bellmanlemma ] , and the supermartingale property again , we obtain @xmath190 \nonumber \\ & \geq \mathbf{e}_x\left [ e^{-(r+(1-p)\lambda)\tau_1}g_a(x_{\tau_1 } )
- e^{-(r+(1-p)\lambda)\tau_1}k \right ] \\ & = \mathbf{e}_x\left [ \int_{\tau_1}^{\sigma_1 } e^{-(r+(1-p)\lambda)t}(h(x_t)-c)dt + e^{-(r+(1-p)\lambda)\sigma_1 } g_i(x_{\sigma_1 } ) - e^{-(r+(1-p)\lambda)\tau_1}k \right ] \nonumber \\ & \geq \mathbf{e}_x\left [ \int_{\tau_1}^{\sigma_1 } e^{-(r+(1-p)\lambda)t}(h(x_t)-c)dt - e^{-(r+(1-p)\lambda)\tau_1}k \right ] + \mathbf{e}_x \left[e^{-(r+(1-p)\lambda)\tau_2 } g_i(x_{\tau_2})\right ] .
\nonumber\end{aligned}\ ] ] by repeating this argument , we find that @xmath191 \\ & + \mathbf{e}_x \left[e^{-(r+(1-p)\lambda)\tau_{j+1 } } g_i(x_{\tau_{j+1}})\right ] \\ & \geq \mathbf{e}_x\left [ \sum_{j=1}^n \int_{\tau_j}^{\sigma_j } e^{-(r+(1-p)\lambda)t}(h(x_t)-c)dt - e^{-(r+(1-p)\lambda)\tau_j}k \right ] , \end{aligned}\ ] ] for all @xmath192 .
let @xmath193 and apply dominated convergence theorem .
then , by taking supremum over all @xmath57 , we obtain @xmath194 for all @xmath152 . to establish the opposite inequality ,
let @xmath195 , where @xmath196 is given by .
we find using remark [ suffiremark ] that in this case , all inequalities in become equalities .
therefore @xmath197 \\ & + \mathbf{e}_x \left[e^{-(r+(1-p)\lambda)\tau^*_{j+1 } } g_i(x_{\tau^*_{j+1}})\right],\end{aligned}\ ] ] for all @xmath154 .
clearly @xmath198 .
this implies that @xmath199 almost surely .
using lemma [ limitlemma ] , we find that @xmath200 \rightarrow 0,\ ] ] as @xmath180 .
letting @xmath193 , we obtain by dominated convergence @xmath201 , \end{aligned}\ ] ] which implies that @xmath202 for all @xmath152 .
thus @xmath203 and the sequence @xmath204 yields the optimal value .
next , we consider case ( 1 )
. it is sufficient to show that for every @xmath57 , we have @xmath205 \leq 0,\ ] ] for all @xmath61 . by the strong markov property of @xmath0 and the memoryless property of exponential distribution , we only need to establish that @xmath206 = ( r_{r+\lambda}\bar{h})(x)\leq k.\ ] ] by the monotonicity of @xmath63 , we find that @xmath207 the proof is now complete . under assumptions [ sa ] , it is possible to prove , similarly to theorem [ mainthrm ] , that the candidate @xmath208 and that the optimal entry threshold is @xmath99 .
the properties of the value functions and the optimal entry threshold with respect to the parameter @xmath8 were studied in detail in @xcite in the case of gbm dynamics . for a general diffusion process ,
a similar analysis is a formidable task and is left for future research .
we illustrate in this section some of our results using explicit examples .
assume that the process @xmath0 follows a geometric brownian motion , that is , a diffusion process with the infinitesimal generator @xmath209 we assume that @xmath210 .
in this case , the process @xmath211 almost surely as @xmath212 .
it is easy to check that for an arbitrary @xmath213 , the functions @xmath214 and @xmath215 read as @xmath216 we verify readily that the densities @xmath217 and @xmath218 read as @xmath219 and @xmath220 .
moreover , the wronskian determinant @xmath221 . using this information and the formula , we find that the optimal entry threshold @xmath99 is characterized by @xmath222 which can be further simplified to @xmath223 this is the expression ( 4.19 ) in @xcite .
as we have observed already in the general case , the optimal entry threshold is independent of the parameter @xmath68 .
we illustrate the effect of the parameter @xmath68 on the value @xmath96 .
let @xmath224 .
straightforward integration yields @xmath225 since @xmath226 for all @xmath227 $ ] , the term @xmath228 becomes smaller as @xmath68 approaches zero .
similarly , we observe that the integral term becomes smaller as @xmath68 approaches zero .
finally , since the wronskian @xmath229 increases as @xmath68 approaches zero , we conclude that as the probability of success @xmath68 becomes smaller , the value of the idle problem decreases on @xmath98 .
let @xmath168 .
straightforward integration and an application of yields @xmath230.\end{aligned}\ ] ] we observe by standard differentiation that in the expression above , both integrands decrease on their respective intervals as @xmath68 approaches zero . summarizing , we conclude that the value @xmath96 decreases on @xmath231 as @xmath68 decreases .
this observation is in line with the general result and illustrates that increased catastrophe risk decreases value . to conclude
, we graphically illustrate the value function @xmath96 for various values of @xmath68 .
let @xmath232 .
it is easy to verify that the optimal entry threshold @xmath233 ^ 2.\ ] ] in figure [ fig1 ] we illustrate the curves @xmath234 and and @xmath235 around the optimal entry threshold @xmath99 under the parameter configuration @xmath236 , @xmath237 , @xmath238 , @xmath239 , and @xmath240 .
the figure suggests that the curves tangent at @xmath99 .
this is in line with the smoothness requirement of @xmath96 . in figure [ fig2 ]
we illustrate the effect of parameter @xmath68 on the value @xmath96 under the parameter configuration @xmath236 , @xmath237 , @xmath238 , and @xmath239 .
the values of @xmath68 are @xmath242 , @xmath243 , @xmath244 , and @xmath245 and the curves are colored such that the hue becomes lighter as the probability @xmath68 decreases .
the figure shows that decreasing @xmath68 decreases value , this is in line with our general result . as a generalization of the geometric brownian motion ,
we consider the diffusion @xmath0 with infinitesimal generator @xmath247 with @xmath12 , @xmath13 , and @xmath248 positive .
this process exhibits mean reversion and has been applied successfully in investment theory , see @xcite .
we point out that when @xmath249 , this process reduces to a geometric brownian motion . a straightforward computation yields the functions @xmath217 and @xmath218 : @xmath250 and @xmath251 for all @xmath152 .
furthermore , it is known from the literature that for an arbitrary @xmath213 , the functions @xmath214 and @xmath215 reads as @xmath252 where @xmath67 and @xmath253 are , respectively , the confluent hypergeometric functions of first and second type , see @xcite . due to the complicated nature of these functions , we resort to numerical solution of the optimal entry threshold @xmath99 and the value function @xmath96 . in figure [ fig3 ]
we illustrate the curves @xmath234 and and @xmath235 around the optimal entry threshold @xmath99 under the parameter configuration @xmath236 , @xmath237 , @xmath238 , @xmath239 , @xmath254 , and @xmath240 .
the figure suggests that the curves tangent at @xmath99 . in figure [ fig4 ]
we illustrate the effect of parameter @xmath68 on the value @xmath96 under the parameter configuration @xmath236 , @xmath237 , @xmath238 , @xmath239 , and @xmath254 .
the values of @xmath68 are @xmath242 , @xmath243 , @xmath244 , and @xmath245 and the curves are colored such that the hue becomes lighter as the probability @xmath68 decreases .
the figure shows that decreasing @xmath68 decreases value , this is in line with our general result . | we study an optimal investment problem with multiple entries and forced exits .
a closed form solution of the optimisation problem is presented for general underlying diffusion dynamics and a general running payoff function in the case when forced exits occur on the jump times of a poisson process .
furthermore , we allow the investment opportunity to be subject to the risk of a catastrophe that can occur at the jumps of the poisson process . more precisely , we attach iid bernoulli trials to the jump times and if the trial fails , no further re - entries are allowed .
we show in the general case that the optimal investment threshold is independent of the success probability is the bernoulli trials .
the results are illustrated with explicit examples .
= 1 |
spin glasses are the type of random magnets in which both ferromagnetic and antiferromagnetic interactions coexist and compete , thereby giving rise to the effects of frustration and quenched randomness .
spin glasses ( sgs ) have been regarded as a typical example of complex systems , and their ordering properties have been studied quite extensively , either experimentally , analytically or numerically @xcite .
experimentally , we now have fairly convincing evidence that sg magnets exhibit an equilibrium phase transition at a finite temperature into the glassy ordered state .
this has been particularly well established in canonical sgs which represent a class of dilute transition - metal alloys soluted in the noble - metal host . in theoretical studies of the ordering of sgs ,
a statistical model called the edwards - anderson ( ea ) model has been widely used @xcite . in this model ,
spins are located on a certain regular lattice , say , on a three - dimensional ( 3d ) simple cubic lattice , and are assumed to interact via the random exchange interaction taking either ferromagnetic or antiferromagnetic sign .
one crucially important ingredient of the ea model is its spin symmetry . in its original version
as proposed by edwards and anderson , the spin was assumed to be the heisenberg one with three spin components .
in fact , this is expected to be a good approximation for most of real sg materials , particularly for canonical sgs which possess relatively weak magnetic anisotropy .
another well - studied sg model is the ising ea model with only one spin component .
although an ideal ising sg is relative scarce in real sg materials , it has been widely used in numerous sg simulations performed in the past because of its simplicity .
a naive expectation might be that the weak magnetic anisotropy inherent to real magnets eventually causes a crossover from the isotropic heisenberg behavior to the anisotropic ising behavior even in real sg materials @xcite , so that the ising ea model corresponding to the strongly anisotropic limit is enough to describe the asymptotic behavior of weakly anisotropic sg magnets .
although such a naive expectation sounds plausible from the standard view of critical phenomena , the actual situation is not necessarily so simple as we shall see below .
obviously , it is also crucially important to fully understand the ordering properties of the heisenberg ea model , in particular , whether the heisenberg ea model behaves similarly or differently from the ising ea model . indeed , via recent extensive studies , it now becomes increasing clear that the 3d heisenberg ea model exhibits an intriguing ordering behavior quite different from that of its ising counterpart . among others , a special degree of freedom
called a `` chirality '' , totally absent in the corresponding ising model , might play a crucial role in the ordering process of the 3d heisenberg ea model @xcite , which might be crucially important in the ordering of real weakly anisotropic sg materials @xcite . in the present article
, we wish to review the recent studies on the ordering properties of the two typical models of sgs , _
i.e. _ , the 3d ising ea model versus the 3d heisenberg ea model , focusing on their similarities and differences
. indeed , although the recent numerical studies have indicated that both the ising ea model and the heisenberg ea model exhibit a finite - temperature transition , the nature of the ordering is often in sharp contrast with each other . in the present article , we wish to focus on the following four issues .
\(a ) is there a thermodynamic transition in zero field ?
\(b ) if so , what are the associated critical properties ?
\(c ) what is the nature of the ordered state ? does it exhibit a replica - symmetry breaking ( rsb ) ? if yes , what type ?
\(d ) is there a thermodynamic transition in applied magnetic fields ? in the following , we shall examine the ordering properties of the 3d ising sg in section 2 , and those of the 3d heisenberg sg in section 3 .
section 4 is devoted to summary and discussion .
in this section , we review the ordering properties of the 3d ising ea model .
the hamiltonian of the 3d ising ea model is given by , @xmath0 where @xmath1 is an ising spin variable located at the @xmath2-th site of a 3d simple - cubic lattice , @xmath3 is an external field intensity , and the @xmath4 sum is taken over all nearest - neighbor pairs on the lattice .
periodic boundary conditions are applied in all directions .
the nearest - neighbor couplings @xmath5 are quenched random variables with zero mean and standard deviation unity . for the form of the @xmath5 distribution , either the gaussian distribution or the binary distribution ( @xmath6 distribution ) is widely used .
the ordering properties of the 3d ising ea model has been studied quite extensively for years as a standard model of sgs since the beginning of the sg research in 1970 s .
note that the model has a finite - range ( nearest - neighbor ) interaction so that , in contrast to the mean - field - type ( or an infinite - ranged ) model such as the sherrington - kirkpatrick ( sk ) model @xcite , the correlation effect is taken into account in the model . in spite of its apparent simplicity , however , the ordering properties of the 3d ising sg model turns out to be highly nontrivial and hard to be elucidated .
so far , rigorous analytical information on this model has been scarce , and most of the information we now have for this model is necessarily a numerical one . nevertheless , after more than 30 years of extensive research , we now have several established ( or at least almost established ) knowledge about the ordering properties of the 3d ising ea model .
we summarize some of them below , together with several open issues . in earlier studies , the issue of whether the 3d ising ea model exhibits a finite - temperature transition or not remained controversial .
monte carlo ( mc ) simulations on the model made by ogielski and by bhatt and young in 1985 changed this situation , giving strong support to the occurrence of a finite - temperature sg transition @xcite . since then , many numerical works have been made , all of which have agreed in that the 3d ea model exhibits a thermodynamic sg transition at a finite temperature @xcite . in particular , the correlation - length ratio and the binder ratio played an important role in identifying the sg transition point . in fig.1 , the correlation - length ratio @xmath7 , defined by the finite - size sg correlation length @xmath8 divided by the linear size of the system @xmath9 , calculated by katzgraber _
et al _ for the @xmath6 ising ea model of @xmath10 is shown , together with the binder ratio calculated by the same authors @xcite .
the correlation - length ratio and the binder ratio are dimensionless quantities so that the data of various @xmath9 should be scale - invariant and exhibits a crossing behavior at the respective sg and chiral - glass transition points . indeed , clear fan - out of the correlation - length ratio and the binder ratio is observed at @xmath11 , indicating the occurrence of a finite - temperature sg transition . )
@xmath12 is the inverse temperature .
taken from [ h. g. katzgraber , m. k@xmath13rner and a. p.
young , phys .
rev . b*73 * ( 2006 ) 224432 ] .
, title="fig : " ] ) coupling .
@xmath12 is the inverse temperature .
taken from [ h. g. katzgraber , m. k@xmath13rner and a. p.
young , phys .
rev . b*73 * ( 2006 ) 224432 ] .
, title="fig : " ] by contrast , reliable estimates of the associated sg exponents are more difficult . concerning the critical - point - decay exponent ( or the anomalous dimension ) @xmath14 ,
various simulations have given nearly a common value of @xmath15 .
by contrast , concerning the correlation - length exponent @xmath16 , smaller values around @xmath17 reported by earlier simulations @xcite have been revised by later simulations to significantly larger values .
indeed , most recent mc simulations have yielded @xmath18 , together with @xmath15 @xcite .
the sg susceptibility exponent @xmath19 is given by @xmath20 .
as an example , we show in fig.2 the finite - size scaling plot of the correlation - length ratio of the @xmath6 ising ea model reported by campbell _
et al _ , which yielded @xmath21 @xcite . )
the exponent @xmath16 is taken to be 2.72 .
taken from [ i. a. campbell , k. hukushima , and h. takayama : phys .
* 97 * ( 2006 ) 117202 ] . ]
although a good experimental realization of the ising sg has been rather scarce , at least one example , an insulating sg magnet femntio@xmath22 , has been thoroughly studied .
indeed , the experimental exponent values measured for femntio@xmath22 turns out to be roughly consistent with the numerical values of the 3d ising sg @xcite .
meanwhile , in many of real sg magnets including canonical sgs , magnetic anisotropy is much weaker than the isotropic exchange interaction so that they might be described as a weakly anisotropic heisenberg - like sg .
it should be stressed that , in contrast to a naive expectation mentioned above , the experimentally determined sg critical exponents of such weakly anisotropic heisenberg - like sgs largely deviate from the exponent values of the 3d ising ea model .
indeed , for typical canonical sg materials , consistent experimental estimates are now available thanks to careful experimental measurements @xcite .
the exponents determined by various authors for canonical sgs indeed come close to each other , yielding the values @xmath23 , @xmath24 , @xmath25 and @xmath26 , which are clearly at odd with the exponent values of the 3d ising ea model .
the experimental @xmath27 and @xmath19 are about half and the sign of @xmath28 is reversed .
to elucidate the origin of the discrepancy observed between the critical properties of experimental heisenberg - like sg magnets and of the 3d ising sg is one of important issues left in the sg research .
one of hot issues in the sg research has been concerned with the nature of the sg ordered state : in particular , whether the sg ordered state spontaneously breaks a replica symmetry or not @xcite .
two typical views have been common .
one is a droplet picture , which claims that the sg ordered state is a `` disguised ferromagnet '' without a spontaneous rsb @xcite .
the other is a hierarchical rsb picture inspired by the exact solution of the mean - field model , which claims that the sg ordered state is intrinsically more complex accompanied with a hierarchical or full rsb where the phase space is hierarchically organized in the sg ordered state @xcite . hot debate has continued over years concerning which view applies to the ordered state of real sg magnets . for more quantitative discussion ,
it is convenient to introduce an `` overlap '' variable @xmath29 , which is defined by @xmath30 where @xmath31 represents the @xmath2-th spin variable of the `` replica '' @xmath32 and the summation is taken over all @xmath33 spins of the system . replicas @xmath32 and @xmath12 mean here the two independent copies of the system with the same realization of quenched randomness .
one can then consider the distribution function of the overlap variable @xmath34 @xmath35,\ ] ] where @xmath36 represents a thermal average and [ @xmath37 an average over the quenched disorder ( configurational average ) .
some typical forms of @xmath34 in the thermodynamic limit is illustrated in fig.3 .
the droplet picture claims that the overlap distribution describing the sg ordered state to be a trivial one in the thermodynamic limit consisting of just two delta functions located at @xmath38 and at @xmath39 .
it means that the sg ordered state consists of unique pure state and its symmetry counterpart , irrespective of its apparent complexity in real - space spin pattern . in the hierarchical rsb picture , by contrast , @xmath34 exhibits a continuous plateau part spanning between the two delta - function peaks at @xmath40 .
it means that the phase space is divided into infinitely many pure states organized in a hierarchical manner , each of which is separated by infinitely high free - energy barrier .
other types of rsb have also been known .
one well - known example might be a one - step rsb , in which @xmath34 possesses a central @xmath41-function peak at @xmath42 in addition to the self - overlap peaks at @xmath40 . in this case
, the phase space is divided into many components , but all of them , except for itself and its symmetry partner , are completely dissimilar or orthogonal .
it is realized , _
e.g. _ , in the ordered states of the mean - field potts glass or of the mean - field @xmath43-spin model @xcite . in the past , such a one - step rsb
has often been discussed in the context of the structural - glass problem rather than sg problem .
the combination of the full rsb and the one - step rsb is also possible in certain models .
many numerical studies have focused on the 3d ea model to solve this issue of rsb , asking whether this model exhibits a hierarchical rsb or no rsb .
the controversy has not yet been solved completely .
however , most of the recent studies have indicated that the overlap distribution @xmath34 persistently exhibits a nontrivial component suggesting the occurrence of a hierarchical - type rsb @xcite . as a typical example , we show in fig.4 the overlap distribution @xmath34 of the 3d gaussian ising ea model of various lattice sizes at a temperature @xmath44 which is well below the estimated sg transition temperature @xmath45 @xcite .
in contrast to the situation in zero field , the question of the existence / nonexistence of an equilibrium transition under external fields has remained quite controversial for years and not yet been completely settled .
in fact , the question is closely related to the one discussed in the previous subsection , _
i.e. _ , that of an rsb .
for example , if the sg ordered state of the 3d ising ea model does not accompany an rsb , the model is likely to exhibit no finite - temperature transition under finite fields since applied fields destroy the global @xmath46 spin - inversion symmetry , a symmetry broken at the zero - field transition . if , on the other hand , the sg ordered state of the model accompanies an rsb , a finite - temperature transition associated with rsb may become possible even in the situation where no global symmetry is left under fields , though it does not immediately necessitate the occurrence of an in - field transition .
although there is no completely consensus among researchers , most of recent numerical studies point to the absence of an equilibrium transition under finite fields @xcite .
such numerical observations seem consistent with the experimental observation on femntio@xmath22 : for this compound , mattsson _ et al _ indicated that there was no in - field phase transition @xcite , in sharp contrast to the zero - field case .
meanwhile , the absence of an equilibrium transition in finite fields seem in apparent contrast to the possible occurrence of a hierarchical - type rsb observed in zero field .
further study is required to clarify this point .
in the present section , we deal with the ordering of the 3d heisenberg ea model .
the hamiltonian of the 3d heisenberg ea model is given by , @xmath47 where @xmath48 is now a three - component unit vector located at the @xmath2-th site of a 3d simple - cubic lattice , @xmath3 is an external field intensity applied along the @xmath49-direction , and the @xmath4 sum is taken over all nearest - neighbor pairs .
periodic boundary conditions are applied in all directions .
the nearest - neighbor couplings @xmath5 are quenched random variables with zero mean and standard deviation unity . as in the ising case , either the gaussian distribution or the binary distribution ( @xmath6 distribution ) is widely used for the form of the @xmath5 distribution .
the ordering properties of the 3d heisenberg ea model have long been studied @xcite , and are hotly debated even now . earlier numerical simulations on the model suggested in common that the model exhibited only a zero - temperature transition @xcite , in apparent contrast to experiments .
common attitude in the community at that time was to invoke the weak magnetic anisotropy inherent to real materials to explain this apparent discrepancy with experiments , assuming that the weak anisotropy caused a rapid crossover from the @xmath50 heisenberg behavior to the @xmath51 ising behavior @xcite .
in fact , however , the situation was not quite satisfactory as already discussed above , in view of the fact that the experimental exponent values measured for canonical sgs are actually far from the ising sg values , and no clear sign of heisenberg - to - ising crossover has been observed in experiments . in 1992 ,
the present author suggested that the 3d heisenberg ea model might exhibit a finite - temperature transition _ in its chiral sector _
chirality is a multispin variable representing the sense or the handedness of the noncollinear or noncoplanar structures induced by frustration , _ i.e. _ , whether the frustration - induced noncollinear or noncoplanar spin structure is right- or left - handed .
the scalar chirality @xmath52 is defined by the product of three neighboring spins by @xmath53 it has subsequently been suggested that , in the ordering of the 3d heisenberg sg , the chirality was `` decoupled '' from the spin , the chiral - glass order taking place at a temperature higher than the sg order , @xmath54 @xcite .
based on such a spin - chirality decoupling picture of the 3d isotropic heisenberg sg , a chirality scenario of experimental sg transition was proposed by the author @xcite : according to this scenario , the chirality is a hidden order parameter of real sg transition .
the chirality scenario consists of the two parts , _
i.e. _ , the `` spin - chirality decoupling '' part for a completely isotropic system and the `` spin - chirality recoupling '' part for a weakly anisotropic system @xcite .
the first part , the spin - chirality decoupling , is a key ingredient of the scenario .
it claims that the fully isotropic 3d heisenberg sg exhibits a peculiar two - step ordering process , in which the systems exhibits , with decreasing the temperature , first the glass ordering of the chirality at a finite temperature @xmath55 spontaneously breaking only a discrete @xmath46 symmetry with preserving the continuous @xmath56 symmetry , and at a lower temperature @xmath57 exhibits the glass ordering of the spin itself fully breaking both the @xmath46 and @xmath56 symmetries .
the higher transition at @xmath55 associated with the discrete @xmath46 symmetry breaking is called the `` chiral - glass transition '' , while the intermediate phase between @xmath58 and @xmath59 where only the @xmath46 symmetry is broken with preserving the continuous @xmath56 symmetry is called the `` chiral - glass state '' .
the second part of the chirality scenario concerns the role of the weak anisotropy which inevitably exists in real sg magnets to certain extent .
the chirality scenario claims that the weak random magnetic anisotropy , which reduces the hamiltonian symmetry from @xmath60 to only @xmath46 , weakly `` mixes '' the chirality to the spin sector , and the chiral - glass transition hidden in the chiral sector in fully isotropic system is ` revealed ' in the spin sector in weakly anisotropic system . in this scenario ,
the chiral - glass transition of the fully isotropic system , not the sg transition of the isotropic system , dictates the sg of real weakly anisotropic sgs .
the chirality scenario is capable of explaining several long - standing puzzles concerning the experimental sg transition in a natural way , _
e.g. _ , the origin of the non - ising critical exponents experimentally observed in canonical sgs , and remains to be an attractive hypothesis in consistently explaining various experimental observations for canonical sgs @xcite . in recent numerical studies of the 3d heisenberg ea model , although consensus now seems to appear that the 3d heisenberg sg indeed exhibits a finite - temperature transition of some sort @xcite , the nature of the transition , especially whether the model really exhibits the spin - chirality decoupling , is still under hot debate @xcite .
thus , the issue of whether the spin - chirality decoupling really occurs in the 3d heisenberg sg has remained controversial in spite of the potential importance in understanding the nature of experimental sg ordering . while several numerical results in favor of the occurrence of the spin - chirality decoupling were reported @xcite , a simultaneous spin and chirality transition without the spin - chirality decoupling
was claimed in other works @xcite .
the recent simulation of ref.@xcite , however , has provided a fairly strong numerical support for the occurrence of the spin - chirality decoupling .
this calculation indicates that the sg transition point @xmath59 is located about 10% @xmath61 15% below the chiral - glass transition point @xmath58 , which we shall briefly describe below . and @xmath62 , the dip temperature @xmath63 and the crossing temperature @xmath64 of @xmath65 .
the inset exhibits a wider range . ] in fig.5 , we show the recent mc data of the correlation - length ratios for the chirality @xmath66 ( a ) , and for the spin @xmath62 ( b ) , of the isotropic 3d heisenberg ea model with random gaussian coupling @xcite .
the system contains total @xmath67 spins with @xmath9 ranging from 6 to 32 . for further details ,
refer to ref.@xcite . as can be seen from the figure , while the chiral @xmath66 curves cross at temperatures which are only weakly @xmath9-dependent , the spin @xmath62 curves cross at progressively lower temperatures as @xmath9 increases . as an other indicator of the transition
, we show in fig.6 the binder ratios for the chirality ( a ) , and for the spin ( b ) .
the binder ratios are also dimensionless , and are expected to exhibit a scale - invariant behavior at the respective chiral - glass and sg transition points . as can be seen from the figure ,
the chiral binder ratio @xmath65 exhibits a negative dip which deepens with increasing @xmath9 .
the data of different @xmath9 cross on the _ negative _ side of @xmath65 unlike the correlation - length ratio .
these features indicate a finite - temperature transition in the chiral sector . to estimate the bulk chiral - glass and sg transition temperatures quantitatively , we plot in fig.7 the crossing temperature of @xmath66 and @xmath62 for pairs of successive @xmath9 values versus @xmath68 , where @xmath69 is a mean of the two sizes , together with the dip temperature @xmath63 and the crossing temperature @xmath64 of the chiral binder ratio @xmath65 .
the data show a near - linear @xmath68-dependence .
the chiral crossing temperatures of @xmath66 and of @xmath65 exhibits a weaker size dependence than the spin crossing temperature , and are extrapolated to @xmath70 .
the spin crossing temperature exhibits a stronger size dependence , which is extrapolated to @xmath71 .
hence , @xmath59 is lower than @xmath58 by about 10% @xmath61 15% .
various chiral - glass exponents have been estimated via the standard finite - size scaling analysis . as an example
, we show in fig.8 the finite - size scaling plot of the chiral correlation - length ratio @xmath66 in which the effect of the leading correction - to - scaling term has been taken into account .
the exponents determined in this way from various physical quantities yield @xmath72 and @xmath73 , _ etc .
_ , which differ significantly from the standard 3d ising sg values , @xmath18 and @xmath15 @xcite .
the results indicate that the chiral - glass transition belongs to a universality class different from that of the 3d ising sg .
possible long - range and/or many - body nature of the chirality - chirality interaction might be the cause of this difference . concerning the critical properties associated with the sg transition , which is likely to be located below the chiral - glass transition , the available information is still rather limited .
the exponent @xmath74 was reported to be @xmath75 in ref.@xcite , while the exponent @xmath76 to be @xmath77 in ref.@xcite . in this subsection
, we discuss the issue of the nature of the chiral - glass ordered state , _
i.e. _ , whether the chiral - glass state exhibits an rsb , and if so , what type . in this connection
, it should be noticed that the form of the chiral binder ratio @xmath65 shown in fig.6 , which exhibits a prominent negative dip , is quite peculiar .
in fact , this form of the binder ratio resembles the one of the system exhibiting a one - step rsb @xcite .
in fig.9 , we show the chiral - overlap distribution @xmath78 in the chiral - glass phase calculated in ref.@xcite for the 3d heisenberg sg model with the binary coupling . here , the chiral overlap @xmath79 is defined by @xmath80 , where ( 1 ) and ( 2 ) indicate two copies ( replicas ) of the system .
the local chirality variable @xmath81 is defined at the @xmath2th site and in the @xmath82-direction ( @xmath83 ) by @xmath84 where @xmath85 is a unit lattice vector along the @xmath82-direction .
the calculated @xmath78 exhibits besides symmetric side peaks located at @xmath86 corresponding to the long - range chiral - glass order , which grow with increasing @xmath9 , it also exhibits a prominent central peak at @xmath87 , which also grows with increasing @xmath9 .
the existence of such a pronounced central peak is a characteristic feature of the system exhibiting a one - step - like rsb , never seen in the ising sg .
the data strongly suggest that the chiral - glass ordered state exhibits a one - step - like rsb @xcite , in quite contrast to the one of the 3d ising sg .
indeed , recent off - equilibrium experiments of the fluctuation - dissipation ratio of the heisenberg sg magnet , thiospinel cdcr@xmath88in@xmath89s@xmath90 , seems consistent with such an expectation @xcite .
mc also indicates that the chiral - glass ordered state is non - self - averaging @xcite . in this subsection
, we discuss the ordering of the 3d heisenberg sg under finite fields @xcite .
when one applies a field to the fully isotropic heisenberg sg , the symmetry of the hamiltonian reduces from @xmath60 in zero field to @xmath91 under fields , where @xmath46 refers to the chiral degeneracy associated with a spin - reflection operation ( solely in spin space , not in real space ) with respect to an arbitrary plane in spin space including the magnetic - field axis , while @xmath92 refers to the continuous degeneracy associated with a spin - rotation operation ( in spin space , not in real space ) around the magnetic - field axis in spin space .
since the @xmath46 chiral symmetry characterized by the sign of the scalar chirality remains under magnetic fields , the chiral - glass transition is expected to persist under magnetic fields .
of course , applied fields change the symmetry , but lower only the continuous part from @xmath56 to @xmath92 .
since the continuous part has already been decoupled from the discrete @xmath46 part , a natural expectation here would be that the chiral - glass transition persists even under fields essentially of the same type as the zero - field one .
in particular , the chiral - glass transition line under fields should be a regular function of the filed intensity @xmath3 .
since there is a trivial @xmath93 symmetry , the chiral - glass transition temperature under fields should behave for weak fields as @xmath94 ( @xmath95 is a constant ) .
in fact , this yields a transition line resembling the so - called gt line of the mean - field model @xcite , @xmath96 , although the origin of the exponent @xmath97 is entirely different : here , 1/2 is just of regular origin , whereas the exponent @xmath97 in the mean - field model can not be regarded as of regular origin .
the expected phase diagram expected from the chirality scenario is sketched in fig.10 .
the ordering associated with the continuous part should also occur under finite fields .
the in - field transition line associated with the continuous @xmath92 symmetry breaking should be a continuation of the @xmath56 breaking sg transition in zero field . since the broken symmetry is different in zero and finite fields , _
i.e. _ , @xmath56 versus @xmath92 , the sg transition line at low fields should exhibit a singular form , @xmath98 .
the exponent @xmath99 is not yet precisely determined , but may roughly be estimated as @xmath100 .
this @xmath92 breaking transition line is also sketched in fig.10 .
in - field ordering properties of the isotropic 3d heisenberg sg were studied by imagawa and the author also by means of mc simulations , which successfully demonstrates some of the feature of the phase diagram mentioned above @xcite .
the obtained magnetic phase diagram is also consistent with the experimental one for canonical sgs @xcite . in fig.11 , we show the magnetic phase diagram of the 3d isotropic heisenberg ea model with binary coupling as determined by mc simulation @xcite .
( note that the sg transition line was not examined in this work ) . another interesting observation from mc
is that the sg ordered state turns out to be quite robust against applied magnetic fields as can be seen from fig.11 @xcite .
it appears to be stable up to fields as large as 25@xmath101 .
this might be understandable if one notices that the coupling between the chirality and magnetic fields might be rather weak , since magnetic fields couple directly to the spin via the zeeman term , only indirectly to the chirality .
in this article , i reviewed the ordering properties of the two representative models of sgs , the 3d ising ea model and the 3d heisenberg ea model .
particular attention has been paid to i ) whether the model exhibits a finite - temperature transition in zero field or not , and if it does , what are the critical properties of the transition ; ( ii ) what is the nature of the ordered state , whether the model exhibits an rsb , and if yes , what type ; ( iii ) whether the model exhibits a finite - temperature transition in finite fields . concerning the zero - field ordering , both models , _
i.e. _ , the 3d ising and heisenberg ea models , exhibit a finite - temperature transition .
a single sg transition occurs in the ising ea model , whereas successive chiral - glass and sg transitions are likely to occur in the heisenberg ea model , though there still remains controversy about the occurrence of such spin - chirality decoupling . concerning the associated critical properties , recent numerical estimates yield @xmath102 and @xmath103 for the 3d ising sg , while , for the chiral - glass critical properties of the 3d heisenberg sg , recent numerical estimates yield @xmath104 and @xmath105 .
for the sg exponents of the 3d heisenberg sg at its sg transition , which is likely to be located below the chiral - glass transition , recent numerical calculation suggests @xmath106 and @xmath107 . one sees from these results that the universality classes associated with the spin ordering of the ising sg , the chiral and the spin orderings of the isotropic heisenberg sg are all different from each other .
numerically observed critical properties of the 3d ising sg are roughly consistent with the experimental values of the ising - like sg magnet fetio@xmath22 .
numerically observed chiral - glass critical properties of the 3d heisenberg sg are close to those experimentally observed for canonical sgs .
this coincidence supports the chirality scenario of refs.@xcite which claims that the sg critical properties of real weakly anisotropic heisenberg - like sg magnets such as canonical sgs should be equal to the chiral - glass ( _ not _ sg ! ) critical properties of the fully isotropic heisenberg sg . concerning the issue of the rsb in the ordered state , numerical simulations on the 3d ising sg point to the occurrence of a hierarchical - type rsb like the one in the infinite - range sk model ,
although the situation has remained somewhat controversial . for the 3d heisenberg sg , recent simulations suggest that the ordered state might exhibit a one - step - like rsb , the one quite different from that of the ising sg or of the mean - field sk model . according to the chirality scenario
, such a one - step - like rsb should be realized in real weakly anisotropic sg magnets .
concerning the in - field ordering properties , recent numerical simulations suggest that the 3d ising ea model does not exhibit a finite - temperature transition in magnetic fields , which seems consistent with the experiment on the ising - like sg magnet femntio@xmath22 .
by contrast , the 3d heisenberg ea model exhibits a finite - temperature transition in magnetic fields associated with its chiral degrees of freedom , consistently with experimental observations for canonical sgs .
thus , the ordering properties of the 3d ising sg and the heisenberg sg are quite different from each other , each of which might provide a useful reference in attacking various problems related to sgs , not only of random and/or frustrated magnets but also of many systems including glassy materials , complex systems and even information problem . the author is thankful to i.a .
campbell , d.x .
viet , k. hukushima , d. imagawa for the collaboration and many useful discussion , and to t. taniguchi , h. yoshino , t. okubo , m. gingras , e. vincent , m. ocio , m. picco and h. takayama for discussion .
this study was supported by grant - in - aid for scientific research on priority areas `` novel states of matter induced by frustration '' ( 19052006 ) .
we thank issp , tokyo university and yitp , kyoto university for providing us with the cpu time . for reviews on spin glasses , see _ e. g. , _ ( a ) binder k and young a p 1986 rev .
* 58 * 801 ; ( b ) fischer k h and hertz j a 1991 _ spin glasses _ cambridge university press ; ( c ) mydosh j a 1993 _ spin glasses _ taylor & francis ; ( d ) young a p ( _ ed . _ ) 2004 _ spin glasses and random fields _ world scientific , singapore ; ( e ) kawashima n and rieger h 2004 in _ frustrated spin systems _ , ed .
diep ( world scientific , singapore ) | brief review is given on recent numerical research of the ordering of two typical models of spin glasses ( sgs ) , the three - dimensional ( 3d ) ising sg and the 3d heisenberg sg models .
particular attention is paid to the questions of whether there is a thermodynamic transition in zero field , what are the associated critical properties , what is the nature of the ordered state , particularly of a possible replica - symmetry breaking , and whether there is a thermodynamic transition in applied fields .
the properties of the two models are contrasted , and possible relation to experiments is discussed . |
supersymmetry ( susy ) @xcite is regarded as one of the most promising fundamental theories to describe physics at arbitrary high energy beyond the standard model . in msugra
@xcite symmetry breaking is achieved via gravitational interactions and the vast susy parameter space is reduced to only five parameters : the common trilinear coupling at the gut scale @xmath1 ; the sign of the higgsino mixing parameter @xmath2 ; the ratio of the higgs vacuum expectation values tan@xmath3 ; the common scalar and gaugino masses at the gut scale @xmath4 and @xmath5 , respectively . at the tevatron ,
the production of squarks ( @xmath6 ) and gluinos ( @xmath7 ) , super - partners of quarks and gluons , constitutes one of the most promising channels because of the strong couplings involved .
the cascade decay of gluinos and squarks into quarks and gluons will result in a final state consisting of several jets plus missing transverse energy ( @xmath8 ) coming from the neutralinos , which leave cdf undetected .
the inclusive search @xcite for @xmath9 and @xmath10 is performed using 2.0 @xmath11 of data collected by the cdf detector in run ii @xcite . a msugra scenario with @xmath12 , @xmath13 and tan@xmath14 is assumed .
the gluino - squark mass plane is scanned via variation of the parameters @xmath4 ( @xmath15 ) and @xmath5 ( @xmath16 ) .
the pythia @xcite monte carlo program with the isasugra implementation is used to predict the susy spectrum at the tev scale .
light flavor @xmath9 masses are considered degenerate , while 2-to-2 processes involving stop ( @xmath17 ) and sbottom ( @xmath18 ) production are excluded to avoid strong theoretical dependence on the mixing in the third generation .
signal samples are normalized to the next - to - leading - order ( nlo ) cross section calculated with prospino @xcite .
the sm backgrounds are dominated by qcd multi - jets processes where @xmath8 comes from partially reconstructed jets in the final state .
other sources of backgrounds are z and w production in association with jets , top and diboson production where @xmath8 is mostly due to the presence of neutrinos or muons in the final state .
the estimation of both signal and background yields relies on mc techniques .
the background model is tested in different control regions defined to enhance a particular background contribution .
a good agreement is always observed between data and sm prediction in all the control regions . for best sensitivity across the @xmath19 plane ,
three different analyses are developed , requiring at least 2 , 3 or 4 jets in the final state .
the rejection of distinct sm backgrounds relies on specific cuts on kinematic variables including the transverse energy of the leading jets , the @xmath8 of the event , and the angle between the @xmath8 and the direction of the jets .
different selection criteria are defined for each final state to enhance the signal over background separation in different regions of the msugra phase space .
a detailed study of the systematic uncertainties is carried out for each final state : the dominant systematic on the signal efficiencies and sm background yields comes from the @xmath20 uncertainty on the jet energy scale ( jes ) .
this translates into a @xmath21 and @xmath22 uncertainty on the signal and background estimation , respectively .
other systematic uncertainties are related to the modeling of the initial- and final - state radiation ( isr / fsr ) and the choice of renormalization scale and pdfs . finally , a @xmath23 uncertainty on the integrated luminosity is also included .
the number of expected sm events is @xmath24 for the 4-jets analysis , @xmath25 for 3-jets and @xmath26 for 2-jets .
the observed events in @xmath27 of data are 45 , 38 and 18 respectively .
no significant deviation from the sm prediction is detected and the results are translated into exclusion limits on gluino and squark production .
the limits are computed with a bayesian approach at @xmath28 confidence level ( cl ) and including all the systematics and their correlations in the calculation .
depending on gluino and squark masses , cross sections in the range @xmath29 to @xmath30 are excluded ( see fig .
[ inclusive_limit ] left ) . in an msugra scenario
for which @xmath31 this analysis excludes masses up to @xmath32 .
squark masses below @xmath33 are excluded in any case ( see fig .
[ inclusive_limit ] right ) . .
_ right _ : observed ( red region ) and expected ( dashed - line ) 95% cl exclusion limits on gluino / squark production in the @xmath19 plane .
regions excluded by previous experiments are shown , together with no - msugra solution region .
[ inclusive_limit],title="fig:",width=302 ] .
_ right _ : observed ( red region ) and expected ( dashed - line ) 95% cl exclusion limits on gluino / squark production in the @xmath19 plane .
regions excluded by previous experiments are shown , together with no - msugra solution region .
[ inclusive_limit],title="fig:",width=302 ]
under the hypothesis @xmath34 , where @xmath35 is the mass of the light flavor squarks , the gluino pair production @xmath36 is one of the dominant susy processes at tevatron . in the framework of the minimal supersymmetric extension of the standard model ( mssm ) , the assumption @xmath37 leads to a branching ratio of @xmath38 for the @xmath39 decay . in this configuration and with the conservation of the r - parity
, the sbottom production through gluino decays has a very distinctive signature with 4 @xmath40-jets and large missing transverse energy ( @xmath8 ) due to neutralinos in the final state .
the present analysis @xcite is based on @xmath41 of data collected with the cdf detector during run ii .
the event selection follows the strategy developed for the inclusive analysis requiring a minimum @xmath8 per event and at least two jets in the final state . to enhance the heavy flavor content , at least one of the two leading jets is required to be @xmath40-tagged .
the spectrum of mssm events is predicted with pythia while prospino is employed to calculate the nlo cross section .
the gluino and the sbottom masses are varied from @xmath42 to @xmath43 , and from @xmath44 to @xmath45 , respectively . the neutralino mass is fixed to @xmath46 and the squark mass to @xmath47 .
the estimation of the dominant background contributions from qcd multi - jet production and light flavor jets mis - tagged as @xmath40-quark comes directly from data .
the remaining sm backgrounds , including the @xmath48 production , @xmath49 and @xmath50 production in association with jets , single top and diboson production , are modeled using pythia .
the sm prediction is tested in background dominated control regions in which signal contribution is negligible .
good agreement between data and sm is always observed .
different sources of systematic uncertainties are studied .
the primary contributions come from the jes ( up to @xmath51 on signal and background ) and the _ mistags _ estimation ( around @xmath52 ) .
final kinematic cuts on the @xmath8 of the event and the energies of the two leading jets are optimized to have the best sensitivity in two different areas of the susy phase space defined according to the value of @xmath53 . in the small @xmath54 signal region
19 events are observed with a sm expectation of @xmath55 . in the large @xmath56 signal region
the expected sm yield is @xmath57 against 25 events observed in the data . since no significant deviation from the sm prediction
was observed , the results are translated into an exclusion limit on sbottom production process .
limit is computed with a bayesian approach at @xmath28 cl and including all the systematic uncertainties and their correlations in the calculation .
cross sections down to @xmath58 are excluded .
results are also translated into the exclusion limit on gluino and squark masses shown in figure [ sbottom_limit ] : sbottom masses up to @xmath59 are excluded for @xmath60 .
distribution in the large @xmath61 signal region .
data are compared to expected sm and signal yields .
_ right _ : expected and observed exclusion limits at @xmath28 cl on @xmath62 plane .
limits from previous version of the analysis performed by cdf and d@xmath63 during run i and run ii@xcite are also shown .
[ sbottom_limit],title="fig:",width=302 ] distribution in the large @xmath61 signal region .
data are compared to expected sm and signal yields .
_ right _ : expected and observed exclusion limits at @xmath28 cl on @xmath62 plane .
limits from previous version of the analysis performed by cdf and d@xmath63 during run i and run ii@xcite are also shown .
[ sbottom_limit],title="fig:",width=302 ] | we present preliminary results on a search for squarks and gluinos in proton - antiproton collisions with a center - of - mass energy of 1.96 tev and based on about @xmath0 of data collected by the cdf detector in the tevatron run ii .
events with multiple jets of hadrons and large missing transverse energy in the final state are studied within the framework of minimal supergravity ( msugra ) and assuming r - parity conservation .
the results are compared to standard model ( sm ) predictions , and limits on gluino and squark masses are extracted .
a specific search for the supersymmetric partner of the bottom quark produced from gluino decays is carried out using a sample of events with missing transverse energy and two or more jets in the final state where at at least one of the two leading jets is b - tagged .
good agreement is found between data and sm predictions , and limits on gluino and sbottom masses are extracted .
= 1 |
magnetic fields from about 40 kg to 1 gg have been detected in about about 50 ( 2% ) of the 2100 known white dwarfs ( mccook & sion 1998 ) .
a list of all currently known magnetic white dwarfs is found in jordan ( 1997 ) .
magnetic fields in the low field range ( @xmath0 mg ) in both hydrogen and helium rich magnetic white dwarfs can be detected by a relatively simple line pattern .
hydrogen line components have been identified in many objects , but until recently , helium was detected unambiguously only in the magnetic white dwarf feige7 . using line data calculated by kemic ( 1974 ) , achilleos et al .
( 1992 ) found that the spectrum could be well reproduced with a dipole of strength 35 mg displaced by 0.15 white dwarf radii from the stellar center .
since the kemic data were calculated by regarding the magnetic field as a small perturbation to the coulomb interaction and are only valid up to about 20 mg , the kemic tables had to be extrapolated for the feige 7 analysis .
the only other definite helium rich objects with fields of about 15 - 20 mg were discovered by reimers et al .
( 1998 ) in the course of the hamburg eso survey .
magnetic white dwarfs with field strengths @xmath1 mg containing hydrogen could be interpreted ever since the numerical calculations of energy level shifts and transition probabilities for bound - bound transitions by groups in tbingen and baton rouge ( forster et al .
1984 ; rsner et al . 1984 ; henry and oconnell 1984 , 1985 ) . with the help of these data
greenstein 1984 and greenstein et al .
1985 could identify the hitherto unidentified features in with stationary components of hydrogen in fields between about 150 and 500 mg . since the magnetic field on the surface of a white dwarf is not homogeneous but e.g. better described by a magnetic dipole
, the variation of the field strength from the pole to the equator ( a factor of two for a pure dipole field ) smears out most of the absorption lines at larger magnetic field strengths .
however , a few of the line components become stationary , i.e. their wavelengths go through maxima or minima as functions of the magnetic field strength .
these stationary components are visible in the spectra of magnetic white dwarfs despite a considerable variation of the field strengths
. about 80% of all known white dwarfs have nearly pure hydrogen atmospheres ( spectral type da ) .
however , since most of the remaining stars have helium rich atmospheres , we would also expect a significant number of magnetic white dwarfs to belong to the spectral type db in which lines are observed .
therefore it could be expected that the few magnetic white dwarfs with unidentified spectral features , that can not be explained by the line data for hydrogen , are helium rich .
the most famous example is gd229 , where swedlund et al .
( 1974 ) , greenstein et al . ( 1974 ) , landstreet & angel ( 1974 ) , liebert ( 1976 ) , greenstein & boksenberg ( 1978 ) , and schmidt et al .
( 1996 ) found strong absorption features in the optical , infrared , and uv . angel ( 1979 ) already proposed that the absorption bands in this star are due to stationary components of hydrogen or helium .
the basic difficulty in calculating energy levels and oscillator strengths for arbitrary field strengths lies in the fact that the coulomb potential has a spherical symmetry whereas the magnetic field induces a cylindrical symmetry .
this prevents a separation of variables and together with the nonlinear character of the interactions makes even the numerical solution a difficult problem .
in particular the intermediate regime ( @xmath2 ) , in which the majority of magnetic white dwarfs is found , is very complicated , since neither the magnetic nor the coulomb interaction dominates and we therefore encounter a so - called nonperturbative regime where none of the interaction terms can be treated by perturbation theory . in the case of the two electron system
the situation is even more difficult , since the number of degrees of freedom is enhanced significantly and the electron - electron interaction introduces a third competitive force which makes a complex system with a rich variability of the spectrum with changing field strength .
the particular challenge of a comparison of atomic data with observational ones lies in the fact that the energies of many excited states have to be known with a high relative accuracy ( usually @xmath3 ) and for a large number of field strengths in order to identify the stationary transitions . until this year
the status of electronic structure calculations on was such ( see braun et al.1998 , jones et al .
1997 , ruder et al.1994 and references therein ) that a conclusive comparison was impossible and the occurence of in the atmosphere of gd229 had to be considered as a pure speculation as it was the case with respect to hydrogen in until 1984 .
therefore , several alternative explanations have been proposed .
engelhardt & bues ( 1995 ) have tried to explain the regular almost periodical structure of the gd229 spectrum by quasi - landau resonances of hydrogen in a magnetic field of 2.5gg .
streicher et al .
( 1987 ) speculated that two of the features could be due to intersections of hydrogen components in a field of about 25 - 60 mg .
now , the calculations performed in heidelberg ( becken & schmelcher 1998 ) have closed that gap and for the first time precise data for a large number of energy states have become available . in this paper we will show that most of the absorption features in the optical and uv spectrum of gd229 can be identified as stationary line transitions of .
the fixed - nucleus electronic hamiltonian of the helium atom in a strong magnetic field oriented along the @xmath4axis reads as follows @xmath5 where the sum includes the one particle terms in the order : field - free kinetic , orbital zeeman , diamagnetic , coulomb attraction and spin zeeman energies .
the last term represents the repulsive electron - electron interaction .
effects of the finite nuclear mass can easily be taken into account by using the corresponding scaling relations ( becken & schmelcher 1998 ) . according to its symmetries the hamiltonian ( 1 ) possesses four independent conserved quantities : the total spin @xmath6 , its @xmath4component @xmath7 , the @xmath4component @xmath8 of the total spatial electronic angular momentum and the total spatial @xmath4parity @xmath9 ( parity is a combined symmetry of the previous ones ) which will be used to label the electronic states in the form @xmath10 where @xmath11 specifies the degree of excitation within a given symmetry subspace .
calculations are performed separately for each subspace with given quantum numbers . since the spin part of the eigenfunctions of the hamiltonian ( 1 ) are the usual spin singlet and triplet functions we address in the following exclusively the method of investigation for the spatial part of the wave functions .
the key ingredient of our method for the ab initio calculation of the electronic structure of atoms in strong magnetic fields is a basis set of one particle functions which takes on the following appearance @xmath12 where @xmath13 with @xmath14 @xmath15 are nonlinear variational parameters ; @xmath16 is the @xmath17 parity of the one particle function .
the basis set ( 2 ) has its precursor in a more general one ( schmelcher & cederbaum 1988 ) which has successfully been applied for the investigation of the electronic structure of small molecules in strong magnetic fields ( kappes & schmelcher 1996 ) . due to the parameters @xmath15
our basis set is extremely flexible which represents a major advantage in the presence of an external magnetic field .
this allows us to accurately describe the changes in the electronic wave functions with changing field strengths .
the values of the variational parameters @xmath18 are determined by a nonlinear optimization procedure which is performed for the spectrum of both hydrogen and he ii in a magnetic field .
typically 150 one particle functions of type ( 2 ) are optimized in order to describe 15 hydrogenic states with a high accuracy .
the optimization procedure has to be accomplished for each field strength separately ( see becken & schmelcher 1998 and references therein ) . the basic idea for solving the time - independent schrdinger equation belonging to the hamiltonian ( 1 ) is then to construct out of the above optimized one particle functions a set of spatial two electron configurations @xmath19 being symmetric ( spin singlet ) or antisymmetric ( spin triplet ) with respect to the interchange of the electronic coordinates and respecting the above - discussed symmetries .
we represent the exact wave functions @xmath20 by a linear combination of two electron configurations @xmath21 . by virtue of the variational principle
we arrive at a generalized eigenvalue problem whose eigenvalues and eigenvectors are excellent approximations to the exact eigenenergies and wave functions .
we remark that all matrix elements of the hamiltonian ( 1 ) with respect to the basis afunctions ( 2 ) can be evaluated analytically resulting in a complicated series of higher transcendental functions whose numerically stable and efficient implementation required a number of different techniques . the typical number of two - electron configurations @xmath19 , i.e. the dimension of the full hamiltonian matrix to be diagonalized , is of the order of 2000 - 4000 depending on the symmetries . on a powerful workstation 10 - 20
h cpu are necessary to build up and diagonalize the hamiltonian matrix for each field strength and symmetry subspace .
extensive calculations have been performed in the whole range @xmath22 , for 30 electronic states the energy could be determined to an accuracy of at least @xmath23 . for vanishing magnetic quantum number
@xmath24 many excited singlet and triplet electronic states with positive and negative @xmath4parity have been determined .
for nonzero angular momentum @xmath25 both singlet and triplet states have been studied for positive @xmath4parity .
this gives us insight into an essential part of the spectrum of the he i atom with varying field strength and allows us in particular to decide whether he i can be present in the atmosphere of magnetic white dwarfs through an extraction of the stationary components of the transitions in the corresponding regime of field strengths .
the strongest evidence that gd229 does not contain hydrogen came from a hst spectrum taken by schmidt et al .
( 1996 ) : there is no evidence for components of which do not vary very much over a very large range of magnetic field strengths .
the @xmath26 transition has a maximum at 1341and should always be visible even at extremely high field if significant amount of hydrogen were present . for the first time it is now possible to compare the spectral features of gd229 directly with the newly calculated wavelength tables for the line components of in a strong magnetic field ( becken & schmelcher 1998 ) .
since oscillator strengths are not yet available we could not directly compare the spectrum with theoretical models ( see e.g. jordan 1992 ) .
therefore we compared the positions of stationary line components with the observed absorption bands . in total about 150 line components
have been calculated of which about 40 show stationary behaviour at different field strengths .
if one looks at the helium line data up to @xmath27 ( corresponding to @xmath28gg ) it is rather striking to see that the majority ( 26 ) of all stationary components lies in the range @xmath29 ( see fig.[gdfig ] ; note that the wavelengths were determined by spline interpolation in a still relatively crude grid ) .
so far only 10 grid points exist for the range of magnetic field strengths plotted in fig.[gdfig ] . in practically all cases the minimum wavelength of the large number of features seen in the optical , ir , and uv spectrum of gd229
can be attributed to the wavelength minima of the helium components ( listed in table[stationaries ] ) .
this explains why the spectrum of gd229 has such an extremely large number of absorption features . moreover , there are no clear contradictions between the predicted positions of absorptions and the observations
. the only shortcoming of the current analysis is the fact that the strongest features at @xmath30 and at about @xmath31 can only be attributed to two line transitions ( @xmath32 , @xmath33 ) .
since the identification at other wavelengths is convincing we believe that additional stationary line components are hidden in the subspaces of bound - bound transitions that have not been calculated yet ( four subspaces @xmath34 , @xmath35 , and transitions with magnetic quantum numbers @xmath36 ) . in principle
the wavelength maxima ( also shown in the figure ) of the @xmath37 and @xmath38 transitions would fit well to the position of the two strongest absorption .
however , the corresponding field strengths of about 50 mg would be much too low compared to the result found for the other features .
such a large spread of magnetic field strengths is also rather improbable since the observed features are relatively sharp .
an alternative interpretation would be that hydrogen is additionally present in the spectrum .
the strong feature at 4124 in the spectrum of ( which has a polar field strength of 320 mg , e.g. jordan 1992 ) caused by the hydrogen transition @xmath39 could also contribute to the strongest absorption feature in gd229 .
however , in this case we would expect several other absorption troughs , e.g. at 5830 ( @xmath40 ) which are not present in gd229 .
this is also compatible with the finding by schmidt et al .
( 1996 ) that no lyman@xmath41 components are present in the uv spectrum .
rccrno . & component & zero field trans . &
+ 1 & @xmath32 & @xmath42 & 4294 @xmath438 @xmath44 338 min + 2 & @xmath45 & @xmath46 & 3260 @xmath43 6 @xmath44 425 min + 3 & @xmath47 & @xmath48 & 2943 @xmath43 10 @xmath44 455 min + 4 & @xmath49 & @xmath50 & 2802 @xmath43 5 @xmath44 477 min + 5 & @xmath51 & @xmath52 & 10692 @xmath43 40 @xmath44 379 min + 6 & @xmath53 & @xmath54 & 7927 @xmath43 40 @xmath44 406 min + 7 & @xmath55 & @xmath56 & 6984 @xmath43 40 @xmath44 415 min + 8 & @xmath57 & @xmath58 & 3183 @xmath43 6 @xmath44 535 min + 9 & @xmath59 & @xmath60 & 2504 @xmath43 3 @xmath44 658 min + 10 & @xmath61 & @xmath62 & 2295 @xmath43 3 @xmath44 689 min + 11 & @xmath63 & @xmath64 & 2200 @xmath43 5 @xmath44 705 min + 12 & @xmath65 & @xmath66 & 9350 @xmath43 6 @xmath44 573 min + 13 & @xmath67 & @xmath68 & 6990 @xmath43 10 @xmath44 643 min + 14 & @xmath69 & @xmath70 & 6184 @xmath43 10 @xmath44 674 min + 15 & @xmath33 & @xmath71 & 5285 @xmath43 3 @xmath44 50 max + 15 & @xmath33 & @xmath71 & 4812 @xmath43 30 @xmath44 258 min + 16 & @xmath72 & @xmath73 & 4125 @xmath43 3 @xmath44 56 max + 16 & @xmath72 & @xmath73 & 3417 @xmath43 10 @xmath44 429 min + 17 & @xmath74 & @xmath75 & 3012 @xmath43 10 @xmath44 466 min + 18 & @xmath76 & @xmath77 & 2825 @xmath43 3 @xmath44 523 min + 19 & @xmath78 & @xmath79 & 8449 @xmath43 150@xmath44 445 min + 20 & @xmath80 & @xmath81 & 7160 @xmath43 20 @xmath44 499 min + 21 & @xmath82 & @xmath83 & 3582 @xmath43 2 @xmath44 415 min + 22 & @xmath84 & @xmath85 & 2615 @xmath43 5 @xmath44 629 min + 23 & @xmath86 & @xmath87 & 2338 @xmath43 5 @xmath44 689 min + 24 & @xmath88 & @xmath89 & 2208 @xmath43 3 @xmath44 689 min + 25 & @xmath90 & @xmath91 & 7413 @xmath43 40 @xmath44 658 min + 26 & @xmath92 & @xmath93 & 6296 @xmath43 40 @xmath44 658 min + [ stationaries ]
for the first time we have been able to prove that gd229 is indeed a helium rich ( db ) white dwarf at a rather high magnetic field strength .
> from the magnetic field strengths corresponding to the extrema of the wavelengths of the line components we conclude that the range of magnetic fields is about 300 - 700 mg , so that it would be compatible with a dipole field with a polar field strength between 600 and 700 mg . with our analysis we could reject previous ideas that the atmosphere is hydrogen rich .
a detailed modelling of the spectrum and polarization of gd229 will be possible as soon as oscillator strengths become available ; this will enable us to determine the magnetic field structure which is impossible by our wavelength analysis . with the new line data for we will also try to interpret other magnetic white dwarfs with unexplained features : e.g. 1211 ( reimers et al .
1994 , jordan 1997 ) , and four other newly discovered objects from the hamburg eso survey ( reimers et al .
1998 ) .
another good candidate for such an analysis is lb11146b for which a convincing analysis of the spectrum with hydrogen line components arrived at an approximate field strength of about 670 mg ( liebert et al .
1993 , glenn et al .
additionally a very strong and broad feature at about 5600 was present in the spectrum which can not be due to hydrogen .
therefore glenn et al . speculated that helium might be responsible .
however , in our current set of line data no stationary line component of was found at about 670 mg .
_ acknowledgements . _ the deutsche studienstiftung and the deutsche forschungsgemeinschaft ( schm 885/5 - 1 & ko 735/7 - 1 ) are gratefully acknowledged for financial support .
we thank susanne friedrich for the careful reading of the manuscript and the referee richard allen for valuable comments .
achilleos n. , wickramasinghe d.t .
, liebert j. , saffer r.a . , grauer a.d . , 1992 ,
apj 396 , 273 angel j.r.p . , 1979 , in white dwarfs and variable degenerate stars , univ . of rochester press , p. 313
angel j.r.p . , 1978 , araa 16 , 487 becken w. , schmelcher p. , 1998
, in prep .
braun m. , schweizer w. , elster h. , 1998 , phys.rev.a 57 , 3739 engelhardt d. , bues i. , 1995 , in white dwarfs , eds .
d. koester and k. werner , lecture notes in physics , springer , berlin , p. 123
forster h. , strupat w. , rsner w. , wunner g. , ruder h. , herold h. , 1984 , j.phys .
, v 17 , 1301 glenn j. , liebert j. , schmidt g.d . , 1994 , pasp , 106 , 722 greenstein j.l . , schmidt m. , searle l. , 1974 , apj , 190 , l27 greenstein j.l . , boksenberg a. 1978 , mnras 185 , 823 greenstein j.l . , 1984 ,
apjl 281 , l47 greenstein j.l . ,
henry r.j.w . , oconnell r.f . , 1985 ,
apjl 289 , l25 henry r.j.w . ,
oconnell r.f . , 1984 , apj , 282 , l97 henry r.j.w . ,
oconnell r.f . , 1985 ,
pasp , 97 , 333 jones m.d . , ortiz g. , ceperley d.m . , 1997 , phys.rev.e 55 , 6202 jordan s. , 1992 , a&a 265 , 570 jordan s. , 1997 , in white dwarfs , eds .
j. isern , m. hernanz , e. garca - berro , kluwer , p. 397
kappes u. , schmelcher p. , 1996
, phys.rev.a 53 , 3869 kemic s.b . , 1974 ,
133 landstreet j.d . ,
angel j.r.p . , 1974 ,
apj 190 , l 25 liebert j. , 1976 , pasp 88 , 490 liebert j. , bergeron p. , schmidt g.d .
, saffer r.a . , 1993 ,
apj 418 , 426 mccook g.p . ,
, 1998 , submitted to apj streicher r. , seifert w. , ruder h. , wunner g. , 1987 , a&a 173 , l15 reimers d. , jordan s. , beckmann v. , christlieb n. , wisotzki l. , 1998 , submitted to a&a rsner w. , wunner g. , herold h. , ruder h. , 1984 , j.phys . , v 17 , 29 ruder h. , wunner g. , herold h. , geyer f. , 1994 atoms in strong magnetic fields , springer verlag schmelcher p. , cederbaum l.s .
, 1988 phys.rev.a 37 , 672 svedlund j.b .
, kemp j.c . , wolstencroft r.d . , 1974 ,
apj 193 , l11 schmidt g.d . ,
allen r.g .
, smith p.s . , liebert j. , 1996 , apj 463 , 320 | the nature of the strong absorption features in the white dwarf gd229 has been a real mystery ever since it was found to be magnetic in 1974 .
all attempts to explain the spectrum by line components of hydrogen failed . with the first sets of newly calculated line data for in a strong magnetic field we could identify most of the absorption structures in the spectrum with stationary line components in a range of magnetic fields between 300 and 700 mg .
this is much lower than previously speculated and is comparable to the highest fields found in hydrogen rich magnetic white dwarfs .
the reason for the large number of spectral features in gd229 is the extreme accumulation of stationary components of transitions in he i in a narrow interval of field strengths . |
one approach to understanding the physical origin of confinement and chiral symmetry breaking is to attempt to capture the essential physics in a much smaller number of degrees of freedom by projecting from a maximal abelian or maximal center gauge . for simplicity , we consider @xmath0 and denote a link variable transformed by a gauge transformation @xmath1 by @xmath2 with @xmath3 .
maximal abelian gauge corresponds to calculating @xmath4_{x,\mu } -1 \bigr\}\end{aligned}\ ] ] and the abelian projection is performed by truncating @xmath5 and @xmath6 and normalizing to obtain a @xmath7 matrix@xcite .
although results are ambiguous because of gribov ambiguities associated with local minima , the projected results generally reproduce the proper string tension .
maximal center gauge corresponds to calculating @xmath8and center projection is performed by truncating @xmath5 , @xmath6 , and @xmath9 and normalizing the resulting @xmath10 variable to @xmath11 .
the properties of center projection have been discussed widely by greensite and collaborators@xcite and reinhardt and collaborators@xcite .
the set of negative links defines a set of center vortices , corresponding to closed lines in three dimensions and closed surfaces in four dimensions .
vortices may also be thought of as dirac strings connecting monopoles . for subsequent reference , it is useful to visualize a vortex sheet generated by a single negative link in the @xmath12 direction from @xmath13 to @xmath14 .
the vortex sheet is the surface of the cube dual to this link , that is , the unit cube in @xmath15 centered at the origin and having @xmath16 . if one cuts this four - dimensional surface at time @xmath17 and observes it in the three - dimensional @xmath18 space , it corresponds to a square loop piercing the center of each of the plaquettes with wilson loop @xmath19 that share the negative-@xmath12 link .
if two adjacent links in the @xmath12 direction are @xmath19 , the cube grows to a rectangular parallelepiped , and in general surfaces can grow by stacking the dual cubes together .
returning to our single cube dual to a @xmath12 link , we note for example that the @xmath20-@xmath21 surface corresponds to a @xmath22-@xmath12 plaquette and thus corresponds to @xmath23 . in general
, the vortex sheet in the @xmath24 plane corresponds to @xmath25 . by suitably choosing the time direction in euclidean space to either be in or perpendicular to the plane of the sheet
, we can think of the vortex sheets as being sheets of magnetic or electric fields .
hence , a single smoothly bending vortex sheet corresponds locally to either @xmath26 or @xmath27 , but can never generate @xmath28 .
the simplest way to generate lumps of @xmath28 , corresponding to the topological charge density one expects from instantons , is to have an intersection of two mutually perpendicular sheets .
for example , a sheet in the @xmath20-@xmath22 plane at fixed @xmath29 with @xmath30 intersecting a sheet in the @xmath12-@xmath21 plane at fixed @xmath31 with @xmath32 yields a point of @xmath28 at @xmath33 .
if the vortices have finite thickness , the resulting lump of @xmath28 will have finite size .
lattice phenomenology provides several indications that the reduced degrees of freedom subsequent to center projection still embody the essence of confinement and chiral symmetry breaking . the contribution of each vortex piercing a wilson loop is @xmath34 , so that a random ensemble of vortices produces confinement and lattice measurements yield the proper string tension and scaling@xcite .
there are several indications that there are underlying physical structures ( `` thick vortices '' ) corresponding to the `` thin vortices '' obtained by projection .
calculations with twisted boundary conditions to prevent decay to the vacuum yield stationary vortex solutions that are exponentially localized in two dimensions and periodic in the other two dimensions@xcite . when the contribution of a lattice gluon configuration to a wilson loop @xmath35 is labeled by the number @xmath36 of projected vortices that would pierce the loop , one observes that @xmath37 and @xmath38 for large loops@xcite .
this suggests that there is a finite size object in the actual configuration that , when enclosed by a large enough loop , behaves just like an idealized `` thin vortex '' .
in addition , at finite temperature , the vortices appear to become aligned in the time direction , so that they pierce spatial loops yielding an area law .
finally , de forcrand and delia@xcite have studied the effect of removing center vortices . for each link
@xmath39 one determines @xmath40 by center projection and defines the modified distribution @xmath41 , which projects to the trivial configuration .
not only does one lose confinement , but in addition @xmath42 and the topological charge @xmath43 .
that @xmath43 is understandable , since @xmath44 must be far from 1 somewhere for nontrivial topology and projection leads to the trivial sector .
that @xmath42 implies via the banks cacher relation that the zero modes and thus all the instantons and antiinstantons must also be removed .
to understand how the modified configuration removes instantons , we have studied the center projection of a single instanton on a lattice .
projection yields a cluster of parallel @xmath19 links in the center of the instanton in one common direction that we will call time .
this breaks the euclidean invariance of the instanton , and yields a vortex sheet that is pure @xmath45 .
the most relevant configurations are a single dual cube , a @xmath46 cube composed of 8 elementary cubes , and a @xmath47 cube composed of 27 elementary cubes .
to see how the modified distribution cancels an instanton , note that there must be a shift in group space of @xmath48 so @xmath49 .
since @xmath50 if we assume that the full shift of @xmath48 occurs in a single plaquette for an instanton of size @xmath51 and assume a spherical sheet of radius @xmath52 , @xmath53 so that @xmath54 when @xmath55 .
for a cube , @xmath56 for @xmath57 .
hence , if the size of the vortex is @xmath58 , it should cancel @xmath59 . the primary technical problem is local minima corresponding to gribov ambiguities .
we start with a cooled instanton on lattices from @xmath60 to @xmath61 , minimize @xmath62\ ] ] successively on each site by matrix diagonalization , and overrelax by applying @xmath63 with @xmath64 to @xmath65 .
the local minimum depends upon @xmath66 .
for example , a regular gauge instanton with @xmath67 on a @xmath60 lattice yields no vortex for @xmath68 and a @xmath46 vortex for @xmath69 .
similarly the initial gauge matters .
a random gauge transformation at @xmath68 yields a @xmath46 vortex whereas the regular gauge does not .
also , going first to lorentz gauge can prevent finding a vortex solution . by exploring a variety of gauges and initial conditions , the @xmath46 vortex emerged as the minimum solution for this case .
as expected from the analytic estimate of @xmath71 , for a vortex size @xmath52 comparable to @xmath51 , center projection removes the instantons .
figure [ jnfig1 ] shows @xmath72 times the topological charge density as a function of distance for a cooled instanton of size @xmath73 , the instanton with a vortex of size @xmath74 removed , and the density after one and two cooling sweeps .
since @xmath75 begins localized on a shell , it takes several cooling sweeps to see that the integral averages to zero .
after more cooling sweeps , the entire density relaxes to zero .
a similar calculation with size @xmath76 cools back to the original instanton with @xmath77 .
with hindsight , we believe this single - instanton example is an unrepresentative special case .
the problem is that the vortex sheet is a pure @xmath45 field .
although when it has the proper area , it can combine with the existing @xmath78 field of the instanton to cancel the original toplogical charge , the vortex sheet by itself has no topological charge and thus fails to represent the essential phyiscs of the original configuration .
as argued above we expect lumps of @xmath79 to arise from intersections of two vortices .
whereas an isolated instanton will not have vortex sheets extending to infinity , it would be natural for vortices to connect neighboring instantons and antiinstantons .
thus , the relevant case to examine is an ensemble of instantons and antiinstantons , and to see whether projection produces intersections with localized instanton - like lumps of @xmath28 . were this to occur , center vortices would succinctly represent the essential physics of both confinement and chiral symmetry breaking . | we study the relation between center vortices and instantons in lattice qcd
. boxedeps |
in high energy physics ( hep ) experiments , people usually need to select some events with specific interest , so called signal events , out of numerous background events for study . in order to increase the ratio of signal to background
, one needs to suppress background events while keeping high signal efficiency . to this end
, some advanced techniques , such as adaboost@xcite , @xmath0-boost@xcite , @xmath0-logitboost@xcite , @xmath0-hingeboost , random forests @xcite etc . , from statistics and computer sciences were introduced for signal and background event separation in the miniboone experiment@xcite at fermilab .
the miniboone experiment is designed to confirm or refute the evidence for @xmath1 oscillations at @xmath2 found by the lsnd experiment@xcite .
it is a crucial experiment which will imply new physics beyond the standard model if the lsnd signal is confirmed .
these techniques are tuned with one sample of monte carlo ( mc ) events , the training sample , and then tested with an independent mc sample , the testing sample .
initial comparisons of these techniques with artificial neural networks ( ann ) using the miniboone mc samples were described previously@xcite .
this work indicated that the method of boosted decision trees is superior to the anns for particle identification ( pid ) using the miniboone mc samples .
further studies show that the boosted decision tree method has not only better event separation , but is also more stable and robust than anns when using mc samples with varying input parameters .
the boosting algorithm is one of the most powerful learning techniques introduced in the past decade@xcite .
the motivation for the boosting algorithm is to design a procedure that combines many `` weak '' classifiers to achieve a final powerful classifier . in the present work numerous trials are made to tune
the boosted decision trees , and comparisons are made for various algorithms . for a large number of discriminant variables ,
several techniques are described to select a set of powerful input variables in order to obtain optimal event separation using boosted decision trees .
furthermore , post - fitting of weights for the trained boosting trees is also investigated to attempt further possible improvement .
this paper is focussed on the boosting tuning .
all results appearing in this paper are relative numbers .
they do not represent the miniboone pid performance ; that performance is continually improving with further algorithm and pid study .
the description of the miniboone reconstruction packages@xcite , the reconstructed variables , the overall and absolute performance of the boosting pid@xcite , the validation of the input variables and the boosting pid variables by comparing various mc and real data samples@xcite will be described in future articles .
boosting algorithms can be applied to any classifier , here they are applied to decision trees .
a schematic of a simple decision tree is shown in figure 1 , s means signal , b means background , terminal nodes called leaves are shown in boxes .
the key issue is to define a criterion that describes the goodness of separation between signal and background in the tree split .
assume the events are weighted with each event having weight @xmath3 .
define the purity of the sample in a node by @xmath4 where @xmath5 is the sum over signal events and @xmath6 is the sum over background events . note that @xmath7 is 0 if the sample is pure signal or pure background . for a given node let @xmath8 where @xmath9 is the number of events on that node .
the criterion chosen is to minimize @xmath10 to determine the increase in quality when a node is split into two nodes , one maximizes @xmath11 at the end , if a leaf has purity greater than 1/2 ( or whatever is set ) , then it is called a signal leaf , otherwise , a background leaf .
events are classified signal ( have score of 1 ) if they land on a signal leaf and background ( have score of -1 ) if they land on a background leaf . the resulting tree is a _
decision tree_. decision trees have been available for some time@xcite .
they are known to be powerful but unstable , i.e. , a small change in the training sample can produce a large change in the tree and the results . combining many decision trees to make a `` majority vote '' , as in the random forests method , can improve the stability somewhat .
however , as will be discussed in section 6 , the performance of the random forests method is significantly worse than the performance of the boosted decision tree method in which the weights of misclassified events are boosted for succeeding trees .
if there are @xmath12 total events in the sample , the weight of each event is initially taken as @xmath13 .
suppose that there are @xmath14 trees and @xmath15 is the index of an individual tree .
let * @xmath16 the set of pid variables for the @xmath17th event .
* @xmath18 if the @xmath17th event is a signal event and @xmath19 if the event is a background event . * @xmath20 the weight of the @xmath17th event . *
@xmath21 if the set of variables for the @xmath17th event lands that event on a signal leaf and @xmath22 if the set of variables for that event lands it on a background leaf .
* @xmath23 if @xmath24 and 0 if @xmath25 .
there are several commonly used algorithms for boosting the weights of the misclassified events in the training sample .
the boosting performance is quite different using various ways to update the event weights .
the first boosting method is called `` adaboost''@xcite or sometimes discrete adaboost .
define for the @xmath15th tree : @xmath26 calculate : @xmath27 @xmath28 is the value used in the standard adaboost method . change the weight of each event @xmath17 , @xmath29 .
@xmath30 renormalize the weights . @xmath31
the score for a given event is @xmath32 which is just the weighted sum of the scores of the individual trees .
a second boosting method is called `` @xmath0-boost '' @xcite , or sometimes `` shrinkage '' . after the @xmath15th tree ,
change the weight of each event @xmath17 , @xmath29 .
@xmath33 where @xmath0 is a constant of the order of 0.01 .
renormalize the weights . @xmath31
the score for a given event is @xmath34 which is the renormalized , but unweighted , sum of the scores over individual trees .
a third boosting method is called `` @xmath0-logitboost '' .
this method is quite similar to @xmath0-boost , but the weights are updated according to : @xmath35 where @xmath36 for the m@xmath37 tree iteration .
a fourth boosting method is called `` @xmath0-hingeboost '' . again
this method is quite similar to @xmath0-boost , but here the weights are updated according to : @xmath38 where @xmath36 for the m@xmath37 tree iterations .
a fifth boosting method is called `` logitboost''@xcite .
let @xmath39 for signal events and @xmath40 for background events .
initial probability estimates are set to @xmath41 for event @xmath17 , where @xmath42 is the set of pid variables .
let : @xmath43 @xmath44 where @xmath45 is the weight of event @xmath17 .
let @xmath46 be the weighted average of @xmath47 over some set of events . instead of the gini criterion , the splitting variable and point to divide the events at a node into two nodes @xmath48 and @xmath49
is determined to minimize @xmath50 the output for tree @xmath15 is @xmath51 for the node onto which event @xmath17 falls .
the total output score is : @xmath52 the probability is updated according to : @xmath53 a sixth boosting method is called `` gentle adaboost''@xcite .
it uses same criterion as described for the logitboost . here
@xmath54 is same as @xmath55 .
the weights are updated according to : @xmath56 for signal events and @xmath57 for background events , where @xmath58 is the weighted purity of the leaf on which event @xmath17 falls .
a seventh boosting method is called `` real adaboost''@xcite .
it is similar to the discrete version of adaboost described in section 3.1 , but the weights and event scores are calculated in different ways .
the event score for event @xmath17 in tree @xmath15 is given by : @xmath59 where @xmath58 is the weighted purity of the leaf on which event @xmath17 falls .
the event weights are updated according to : @xmath60 and then renormalized so that the total weight is one .
the total output score including all of the trees is given by @xmath61
miniboone mc samples from the february 2004 baseline mc were used to tune some parameters of the boosted decision trees .
there are 88233 intrinsic @xmath62 signal events and 162657 @xmath63 background events .
20000 signal events and 30000 background events were selected randomly for the training sample and the rest of the events were the test sample .
the number of input variables for boosting training is 52 .
the relative ratio is defined as the background efficiency divided by the corresponding signal efficiency and rescaled by a constant value .
the left plot of figure 2 shows the relative ratio versus the signal efficiency for adaboost with 45 leaves per decision tree and various @xmath64 values for 1000 tree iterations .
the boosting performances slightly depend on the @xmath64 values .
adaboost with @xmath28 works slightly better in the high signal efficiency region ( eff @xmath65 65% ) but worse in the low signal efficiency region ( eff @xmath66 60% ) than adaboost with smaller @xmath64 values , 0.8 , 0.5 or 0.3 . to balance the overall performance , @xmath67 is selected to replace the standard value 1 for the adaboost training .
the right plot of figure 2 shows the relative ratio versus the signal efficiency for adaboost with @xmath67 and 1000 tree iterations for various decision tree sizes ranging from 8 leaves to 100 leaves .
adaboost with a large tree size worked significantly better than adaboost with a small tree size , 8 leaves ; the latter number has been recommended in some statistics literature@xcite . typically , it takes more tree iterations for the smaller tree size to reach optimal performance . for this application ,
even with more tree iterations ( 10000 trees ) , results from boosting with small tree size ( 8 leaves ) are still significantly worse ( @xmath6810%-20% ) than results obtained with large tree size ( 45 leaves ) . here , 45 leaves per decision tree is selected ( this number is quite close to the number of input variables , 52 , for the boosting training . ) how many decision trees are sufficient ?
it depends on the mc samples for boosting training and testing . for the given set of boosting parameters selected above
, we ran boosting with 1000 tree iterations .
the left plot of figure 3 shows the relative ratio versus the signal efficiency for adaboost with tree iterations of 100 , 200 , 500 , 800 and 1000 , respectively . the boosting performance becomes better with more tree iterations
. the right plot of figure 3 shows the relative ratio versus the number of decision trees for signal efficiencies of 50% , 60% and 70% which cover the regions of greatest interest for the miniboone experiment .
typically , the boosting performance for low signal efficiencies converges after few hundred tree iterations and is then stable . for high signal efficiency , boosting performance continues to improve as the number of decision trees is increased .
for these particular mc samples , the boosting performance is close to optimal after 1000 tree iterations . for the sake of comparison , the adaboost performance of the boosting training mc samples is also shown in the right plot of figure 3 .
the relative ratios drop quickly down to zero ( zero means no background events left after selection for a given signal efficiency ) within 100 tree iterations for 50%-70% signal efficiencies .
the adaboost outputs for the training mc sample and for the testing mc sample for @xmath69 1 , 100 , 500 and 1000 are shown in figure 4 . the signal and background separation for the training sample becomes better as the number of tree iterations increases .
the signal and background events are completely distinguished after about 500 tree iterations . for the testing samples , however , the signal and background separations are quite stable after a few hundred tree iterations .
the corresponding relative ratios are stable for given signal efficiencies as shown in right plot of figure 3 .
the tuning parameter for @xmath0-boost is @xmath0 .
the left plot of figure 5 shows the relative ratio versus the signal efficiency for @xmath0-boost with @xmath0 values of 0.005 , 0.01 , 0.02 , 0.04 , respectively .
@xmath0-boost with fairly large @xmath0 values for 45 leaves per decision tree and 1000 tree iterations has better performance for the high signal efficiency region ( eff @xmath65 50% ) .
the results from adaboost with @xmath64=0.5 are comparable to those from @xmath0-boost .
@xmath0-boost with @xmath70 0.01 works slightly better because @xmath0-boost converges more quickly with larger @xmath0 values .
however , with more tree iterations , the final performances for different @xmath0 values are very comparable . here
@xmath0 = 0.01 is chosen for further comparisons .
the right plot of figure 5 shows the relative ratio versus the signal efficiency for adaboost and @xmath0-boost using two different ways to split tree nodes .
one way is to maximize the criterion based on the gini index to select the next tree split , the other way is to split the left tree node first . for adaboost , the performance for the `` left node first '' method gets worse for signal efficiency less than about 65% . at about the same signal efficiency , the performance for the two @xmath0-boosts are quite comparable and are comparable with adaboost based on the gini index . however , the @xmath0-boost method based on the gini index becomes worse than the others for high signal efficiency .
larger @xmath0 makes @xmath0-boost converge more quickly , but increasing the the size of decision trees also makes @xmath0-boost converge more quickly .
the performance comparison of adaboost with different tree sizes shown in the right plot of figure 2 is for the same number of tree iterations ( 1000 ) . to make a fair comparison for the boosting performance with different tree sizes , it is better to let them have a similar number of total tree leaves .
the top left plot of the figure 6 shows the relative ratio versus the signal efficiency for adaboost and @xmath0-boost with similar numbers of the total tree leaves , 1800 tree iterations for 45 leaves per tree and 10000 tree iterations for 8 leaves per tree . for a small decision tree size of 8 leaves , the performance of the @xmath0-boost is better than that of adaboost for 10000 trees . for a large decision tree size of 45 leaves , @xmath0-boost has slightly better performance than adaboost at low @xmath62 signal efficiency ( @xmath6665% ) , but worse at high @xmath62 signal efficiency ( @xmath6570% ) .
the comparison between small tree size ( 8 leaves ) and large tree size ( 45 leaves ) with comparable overall decision tree leaves indicates that large tree size with 45 leaves yields @xmath6810%-20% better performance for the miniboone monte carlo samples .
the other five plots in figure 6 show the relative ratio versus the number of tree iterations for adaboost and @xmath0-boost with 45 leaves and 8 leaves assuming signal efficiencies of 40% , 50% , 60% , 70% , 80% , respectively .
the maximum number of tree iterations is 5000 for the large tree size of 45 leaves and 10000 for the small tree size of 8 leaves .
usually , the performance of the boosting method becomes better with more tree iterations in the beginning ; then at some point , it may reach an optimal value and gradually get worse with increasing number of trees , especially in the low signal efficiency region .
the turning point of the boosting performance depends on the signal efficiency and mc samples used for training and test .
generally , if the number of weighted signal events is larger than the number of weighted background events in a given leaf , it is called a signal leaf , otherwise , a background leaf . here ,
the threshold value for signal purity is 50% for the leaf to be called a signal leaf .
this threshold value can be modified , say , to 30% , 40% , 45% , 60% or 70% .
it is seen in figure 7 that the performance of boosted decision trees with adaboost degrades for threshold values away from the central value of 50% . especially for threshold values away from 50% , the @xmath71 of m@xmath37 tree often converges to 0.5 within about 100 tree iterations ; after that the weights of the misclassified events do not successfully update because @xmath72 if @xmath73 . then @xmath74 remains the same as for the previous tree .
typically , the @xmath71 value increases for the first 100 - 200 tree iterations and then remains stable for further tree iterations , causing the weight of m@xmath37 tree , @xmath75 , to decrease for the first 100 - 200 tree iterations and then remain stable . for practical use of the adaboost algorithm , a lower limit , say , 0.01 , on @xmath75 will avoid the impotence of the succeeding boosted decision trees .
this problem is unlikely to happen for @xmath0-boost because the weights of misclassified events are always updated by the same factor , @xmath76 . if differing purity threshold values are applied to boosted decision trees with @xmath0-boost , the performance peaks around 50% and slightly worsens , typically within 5% , for other values ranging from 30% to 70% . the unweighted misclassified event rate , weighted misclassified event rate @xmath71 and @xmath75 for the boosted decision trees with the adaboost algorithm versus the number of tree iterations
are shown in figure 8 , for a signal purity threshold value of 50% . from this plot ,
it is clear that , after a few hundred tree iterations , an individual boosted decision tree has a very weak discriminant power ( i.e. , is a `` weak '' classifier ) .
the @xmath71 is about 0.4 - 0.45 , corresponding to @xmath75 of around 0.2 - 0.1 .
the unweighted event discrimination of an individual tree is even worse , as is also seen in figure 8 .
boosted decision trees focus on the misclassified events which usually have high weights after hundreds of tree iterations .
the advantage of the boosted decision trees is that the method combines all decision trees , `` weak '' classifiers , to make a powerful classifier as stated in the introduction section . when the weights of misclassified events are increased ( boosted )
, some events which are very difficult correctly classify obtain large event weights . in principle , some outliers which have large event weights may degrade the boosting performance . to avoid this effect ,
it might be useful to set an upper limit for the event weights to trim some outliers .
it is found that setting a weight limit does nt improve the boosting performance , and , in fact , may degrade the boosting performance slightly .
however , the effect was observed to be within one standard deviation for the statistical error .
one might also trim events with very low weights which can be correctly classified easily to provide a better chance for difficult events .
no apparent improvement or degradation was observed considering the statistical error .
these results may indicate that the boosted decision trees have the ability to deal with outliers quite well and to focus on the events located around the boundary regions where it is difficult to correctly distinguish signal and background events .
besides adaboost and @xmath0-boost , there are other algorithms such as @xmath0-logitboost , and @xmath0-hingeboost which use different ways of updating the event weights for the misclassified events .
the four plots of figure 9 show the relative ratio versus the signal efficiency for various boostings with different tree sizes .
the top left , top right , bottom left , and bottom right plots are for 500 , 1000 , 2000 , and 3000 tree iterations , respectively . boosting with a large tree size of 45 leaves
is seen to work better than boosting with a small tree size of 8 leaves as noted above .
adaboost and @xmath0-boost have comparable performance , slightly better than that of @xmath0-logitboost .
@xmath0-hingeboost is the worst among these four boosting algorithms , especially for the low signal efficiency region .
the top left , top right , bottom left and bottom right plots of figure 10 show the relative ratio versus the signal efficiency with 45 leaves of @xmath0-boost , adaboost , @xmath0-logitboost and @xmath0-hingeboost for varying numbers of tree iterations . generally , boosting performance continuously improves with an increase in the number of tree iterations until an optimum point is reached . from the two top plots , it is apparent that @xmath0-boost converges more slowly than does adaboost
; however , with about 1000 tree iterations , their performances are very comparable .
there is only marginal improvement beyond 1000 tree iterations for high signal efficiency , and the performance may get worse for the low signal efficiency region if the boosting is over - trained ( goes beyond the optimal performance range ) .
similar plots for the four boosting algorithms with 8 leaves per decision tree are shown in the figure 11 .
results for @xmath0-hingeboost with 30 and 8 tree leaves are shown in the bottom right plots of figures 10 and 11 .
the performance for 200 tree iterations seems worse than that for 100 tree iterations .
this may indicate that its performance is unstable in the first few hundred tree iterations , but works well after about 500 tree iterations .
however , the overall performance of @xmath0-hinge boost is the worst among the four boosting algorithms described above . for some purposes , logitboost has been found to be superior to other algorithms@xcite .
for the miniboone data , it was found to have about 10%-20% worse background contamination for a fixed signal efficiency than the regular adaboost .
logitboost converged very rapidly after less than 200 trees and the contamination ratio got worse past that point .
a modification of logitboost was tried in which the convergence was slowed by taking @xmath77 , the extra factor of @xmath78 slowing the weighting update rate .
this indeed improved the performance considerably , but the results were still slightly worse than obtained with adaboost or @xmath0-boost for a tree size of 45 leaves .
the convergence to an optimum point still took fewer than 300 trees , which was less than the number needed with adaboost or @xmath0-boost .
gentle adaboost and real adaboost were also tried ; both of them were found slightly worse than the discrete adaboost .
relative error ratio versus signal efficiency for various boosting algorithms are listed in table.[table : ratio ] .
.[table : ratio ] relative error ratio versus signal efficiency for various boosting algorithms for miniboone data .
differences up to about 0.03 are largely statistical .
b=0.5 means smooth scoring function described in section 9 . [
cols="^,^,^,^,^,^,^,^",options="header " , ]
the random forests is another algorithm which uses a `` majority vote '' to improve the stability of the decision trees .
the training events are selected randomly with or without replacement .
typically , one half or one third of the training events are selected for each decision tree training .
the input variables can also be selected randomly for determining the tree splitters .
there is no event weight update for the misclassified events . for the adaboost algorithm ,
each tree is built using the results of the previous tree ; for the random forests algorithm , each tree is independent of the other trees .
figure 12 shows a comparison between random forests of different tree sizes and adaboost , both with 1000 tree iterations .
large tree size is preferred for the random forests ( the original random forests method lets each tree develop fully until all tree leaves are pure signal or background ) . in this study
a fixed number of tree leaves were used .
the performance of the random forests algorithm with 200 or 400 tree leaves is about equal . compared with adaboost ,
the performance of the random forests method is significantly worse .
the main reason for the inefficiency is that there is no event weight update for the misclassified events .
one of main advantages for the boosting algorithm is that the weights of misclassified events are boosted which makes it possible for them to be correctly classified in succeeding tree iterations .
considering this advantage , an event weight update algorithm ( adaboost ) was used to boost the random forests .
the performances of the boosted random forests algorithm are then significantly better than those of the original random forests as can be seen in figure 12 .
the performance of the adaboost with 100 leaves per decision tree is slightly better than that of the boosted random forests .
other tests were made using one half training events selected randomly for each tree together with 30% , 50% , 80% or 100% of the input variables selected randomly for each tree split .
the performances of the boosted random forests method using the adaboost algorithm are very stable .
the boosted random forests only uses one half or one third of the training events selected randomly for each tree and also only a fraction of the input variables for each tree split , selected randomly ; this method has the advantage that it can run faster than regular adaboost while providing similar performance .
in addition , it may also help to avoid over - training since the training events are selected partly and randomly for each decision tree .
some recent papers @xcite indicate that post - fitting of the trained boosted decision trees may help to make further improvement .
one possibility is that a selected ensemble of many decision trees could be better than the ensemble of all trees . here
post - fitting of the weights of decision trees was tried .
the basic idea is to optimize the boosting performance by retuning the weights of the decision trees or even removing some of them by setting them to have 0 weight .
a genetic algorithm @xcite is used to optimize the weights of all trained decision trees .
a new mc sample is used for this purpose .
the mc sample is split into three subsamples , mc1 , mc2 and mc3 , each subsample having about 26700 signal events and 21000 background events .
mc1 is used to train adaboost with 1000 decision trees . the background efficiency for mc1 , mc2 and mc3 for a signal efficiency of 60%
are 0.12% , 5.15% and 4.94% , respectively .
if mc1 is used for post - fitting , then the corresponding background efficiency can be driven down to 0.05% , but the background efficiency for test sample mc3 is about 5.5% .
it has become worse after post - fitting .
it seems that it is not good to use same sample for the boosting training and post - fitting .
if mc2 is used for post - fitting , then the background efficiency goes down to 4.21% for the mc2 , and 4.76% for the testing sample mc3 .
the relative improvement is about 3.6% and the statistical error for the background events is about 3.2% .
suppose the mc samples for post - fitting and testing are exchanged , mc3 is used for post - fitting while mc2 is used for testing .
the background efficiency is 4.38% for training sample mc3 and 5.06% for the testing sample mc2 .
the relative improvement is about 1.5% .
a second post - fitting program was tried , the pathseeker program of j.h .
friedman and b.e .
popescu@xcite , a robust regularized linear regression and classification method .
this program produced no overall improvement , with perhaps a marginal 4% improvement for 50% signal efficiency .
it seems that post - fitting makes only a marginal improvement based on our studies .
one of the major advantages of the boosted decision tree algorithm is that it can handle large numbers of input variables as was pointed out previously@xcite . generally speaking ,
more input variables cover more information which may help to improve signal and background event separation .
often one can reconstruct several hundreds or even thousands of variables which have some discriminant power to separate signal and background events .
some of them are superior to others , and some variables may have correlations with others .
too many variables , some of which are `` noise '' variables , wo nt improve but may degrade the boosting performance .
it is useful to select the most useful variables for boosting training to maximize the performance .
new mc samples were generated with 182 reconstructed variables . in order to select the most powerful variables ,
all 182 variables were used as input to boosted decision trees running 150 tree iterations .
then the effectiveness of the input variables was rated based on how many times each variable was used as a tree splitter .
the first variable in the sorted list was regarded as the most useful variable for boosting training .
the first 100 sorted input variables were selected to train adaboost with @xmath67 , 45 leaves per decision tree and 1000 tree iterations .
the dependence of the number of times a variable is used as a tree splitter versus the number of tree iterations is shown for some selected input variables ( variables number 1 , 5 , 10 , 20 , 50 , 80 and 100 ) in the top left plot with linear scale and in the top right plot with log scale . in this way ,
the first 30 , 40 , 60 , 80 , 100 , 120 and 140 input variables were selected from the sorted list to train boosted decision trees with 1000 tree iterations .
a comparison of their performance is shown in the left plot of figure 14 .
the boosting performance steadily improves with more input variables until about 100 to 120 . adding further input variables
does nt improve and may degrade the boosting performance .
the main reason for the degradation is that there is no further useful information in the additional input variables and these variables can be treated as `` noise '' variables for the boosting training .
however , if the additional variables include some new information which is not included in the other variables , they should help to improve the boosting performance . so far only one way to sort the input variables has been described .
some other ways can also be used and work reasonably well as shown in the right plot of figure 14 .
list1 means the input variables are sorted based on the how many times they were used as tree splitters for 150 tree iterations , list2 means the input variables are sorted based on their gini index contributions for 150 tree iterations , and list3 means the variables are sorted according to which variables are used earlier than others as tree splitters for 150 tree iterations .
list1001 , list1002 and list1003 are similar to list1 , list2 and list3 , but use 1000 tree iterations .
the first 100 input variables from the sorted lists are used for boosting training with 1000 tree iterations .
the performances are comparable for 100 input variables sorted in different ways .
however , the boosting performances for list1 and list3 are slightly better than the others .
if an equal number of input variables of 100 are selected from each list , the number of variables which overlap typically varies from about 70 to 90 for the different lists . in other words ,
about 10 to 30 input variables are different among the various lists . in spite of these differences ,
the boosting performances are still comparable and stable . further studies with mc samples generated using varied mc input parameters corresponding to systematic errors show that the boosting outputs are very stable even though some input variables vary quite a lot .
if these same varied mc samples are applied to the anns , it turns out that boosted decision trees work significantly better than the anns for both event separation performance and for stability .
in the standard boost , the score for an event from an individual tree is a simple square wave depending on the purity of the leaf on which the event lands .
if the purity is greater than 0.5 , the score is 1 and otherwise it is @xmath79 .
one can ask whether a smoother function of the purity might be more appropriate .
if the purity of a leaf is 0.51 , should the score be the same as if the purity were 0.99 ?
two possible alternative scores were tested .
let @xmath80purity@xmath79 .
@xmath81 @xmath82 where @xmath83 and @xmath84 are parameters .
tests were run for various parameter values for scores a and b and compared with the standard step function .
performance comparisons of adaboost for various parameters @xmath83 ( left ) and @xmath84 ( right ) values are shown in figure 15 . for a smooth function with @xmath85
, boosting performance converges faster than the original adaboost algorithm for the first few hundred decision trees , as shown in figure 16 .
however , no evidence was found that the optimum was reached any sooner by the smooth function .
the reason is that the smooth function of the purity describes the probability of a given event to be signal or background in more detail than the step function used in the original adaboost algorithm . with an increase in the number of tree iterations
, however , the `` majority vote '' plays the most important role for the event separation .
the ultimate performance of the smooth function with @xmath85 is comparable to the performance of the standard adaboost .
in miniboone , one is trying to improve the signal to background ratio by more than a factor of 100 .
one might expect that one should start by giving the background a greater total weight than the signal .
in fact , giving the background two to five times the weight of the signal slightly degraded the performance . giving the background 0.5 to 0.2 of the weight of the signal gave the same performance as equal initial weights .
for one set of monte carlo runs the pid variables were carefully modified to be flat as functions of the energy of the event and the event location within the detector .
this decreased the correlations between the pid variables .
the performance of these corrected variables was compared with the performance of the uncorrected variables . as expected ,
the convergence was much faster at first for the corrected variable boost .
however , as the number of trees increased , the performance of the uncorrected variable boost caught up with the other . for 1000 trees ,
the performance of the two boost tests was about the same . over the long run ,
boost is able to compensate for correlations and dependencies , but the number of trees for convergence can be considerably shortened by making the pid variables independent . the number of mc events used to train the boosting effectively is also an important issue we have investigated .
generally , more training events are preferred , but it is impractical to generate unlimited mc events for training .
the performance of adaboost with 1000 tree iterations , 45 tree leaves per tree using various number of background events ranging from 10000 to 60000 for training are shown in figure 17 , where the number of signal events is fixed 20000 .
for the miniboone data , the use of 30000 or more background events works fairly well ; fewer background events for training degrades the performance .
pid input variables obtained using the event reconstruction programs for the miniboone experiment were used to train boosted decision trees for signal and background event separation .
numerous trials were made to tune the boosted decision trees .
based on the performance comparison of various algorithms , decision trees with the adaboost or the @xmath0-boost algorithms are superior to the others .
the major advantages of boosted decision trees include their stability , their ability to handle large number of input variables , and their use of boosted weights for misclassified events to give these events a better chance to be correctly classified in succeeding trees .
boosting is a rugged classification method .
if one provides sufficient training variables and sufficient leaves for the tree , it appears that it will , eventually , converge to close to an optimum value .
this assumes that @xmath0 for @xmath0-boost or @xmath64 for adaboost are not set too large .
there are modifications of the basic boosting procedure which can speed up the convergence .
use of a smooth scoring function improves initial convergence . in the last section
, it was seen that removing correlations of the input pid variables improved convergence speed .
for some applications , the use of a boosted natural forests technique may also speed the convergence . for a large set of discriminant variables ,
several techniques can be used to select a set of powerful input variables to use for training boosted decision trees .
post - fitting of the boosted decision trees makes only a marginal improvement in the tests presented here .
we wish to express our gratitude to the miniboone collaboration for the excellent work on the monte carlo simulation and the software package for physics analysis .
this work is supported by the department of energy and the national science foundation of the united states .
j. friedman , _ greedy function approximation : a gradient boosting machine _ , annals of statistics , 29(5 ) , 1189 - 1232(2001 ) ; j. friedman , t. hastie , r. tibshirani , _ additive logistic regression : a statistical view of boosting _ , annals of statistics , 28(2 ) , 337 - 407(2000 ) b.p .
roe , h.j .
yang , j. zhu , y. liu , i. stancu , g. mcgregor , _ boosted decision trees as an alternative to artificial neural network for particle identification _ , nuclear instruments and methods in physics research a , 543 ( 2005 ) 577 - 584 , physics/0408124 .
l. breiman , j.h .
friedman , r.a .
olshen , and c.j .
stone , _ classification and regression trees _
, wadsworth international group , belmont , california ( 1984 ) .
schapire , _ the boosting approach to machine learning : an overview _ , msri workshop on nonlinear estimation and classification , ( 2002 ) . y. freund and r.e .
schapire , _ a short introduction to boosting _ ,
journal of japanese society for artificial intelligence , 14(5 ) , 771 - 780 , ( september , 1999 ) .
( appearing in japanese , translation by naoki abe . )
t. hastie , r. tibshirani , j. friedman , _ elements of statistical learning , data mining , inference and prediction _ , chapter 10 , section 11 , page 324 , springer , ( 2001 ) .
m. dettling , p. buhlmann , _ boosting for tumor classification with gene expression data _ , bioinformatics , vol.19 , no.9 , pp1061 - 1069 , ( 2003 ) z.h .
zhou , j.x .
wu , w. tang , _ ensembling neural networks : many could be better than all _ , artificial intelligence , 137(1 - 2):239 - 263,(2002 ) ; z.h .
zhou , w. tang , _ selective ensemble of decision trees _
, nanjing university , ( 2003 ) . | boosted decision trees are applied to particle identification in the miniboone experiment operated at fermi national accelerator laboratory ( fermilab ) for neutrino oscillations .
numerous attempts are made to tune the boosted decision trees , to compare performance of various boosting algorithms , and to select input variables for optimal performance .
* * studies of boosted decision trees * * for miniboone particle identification |
many successes of modern solid - state physics and chemistry are related with the development of the hohenberg - kohn - sham density - functional theory ( dft),@xcite which is designed for the ground state and based on the minimization of the total energy @xmath5 $ ] with respect to the electron density @xmath6 . for practical applications , dft resorts to iterative solution of single - particle kohn - sham ( ks ) equations , @xmath7 together with the equation for the electron density : @xmath8 defined in terms of eigenfunctions ( @xmath9 ) , eigenvalues ( @xmath10 ) , and the occupation numbers ( @xmath11 ) of ks quasiparticles .
different terms in eq.([eqn : ks ] ) are correspondingly the kinetic - energy operator , the hartree potential , the exchange - correlation potential , and the external potential . in the following
we will also reserve the notation @xmath12@xmath13@xmath14@xmath15@xmath16@xmath15@xmath17@xmath15@xmath18 for the total ks hamiltonian in the real ( @xmath19 ) space .
the exchange - correlation potential is typically treated in the local - density approximation ( lda ) .
it employs an analytical expression for @xmath20 $ ] borrowed from the theory of homogeneous electron gas in which the density of the electron gas is replaced by the local density of the real system .
lda is far from being perfect and there are many examples of so - called strongly - correlated materials for which the conventional lda appears to be insufficient both for the excited - state and ground - state properties.@xcite a typical situation realized in transition - metal ( tm ) oxides is shown in fig .
[ fig.dossummary ] .
we would like to emphasize two points .
+ ( i ) the common feature of many tm oxides is the existence of the well isolated narrow band ( or the group of bands ) located near the fermi level and well isolated from the rest of electronic stares . for compounds shown in fig .
[ fig.dossummary ] , this is the tm @xmath3 band , which is sandwiched between o(@xmath21 ) band ( from below ) and a group of bands ( from above ) , which have an appreciable weight of the tm @xmath22 states ( the meaning of many notations will become clear in sec . [
sec : kineticapplications ] , where we will discuss details of the crystal and electronic structure for the considered oxide compounds ) .
electronic and magnetic properties of these compounds are predetermined mainly by the behavior of this @xmath3 band .
the effect of other bands can be included indirectly , through the renormalization of interaction parameters in the @xmath3 band .
+ ( ii ) the lda description appears to be especially bad for the @xmath3 states located near the fermi level .
it often fails to reproduce the insulating behavior of these compounds , as well as the correct magnetic ground state , which is directly related with the existence of the band gap.@xcite the source of the problem is know to be the on - site coulomb correlations , whose form is greatly oversimplified in the model of homogeneous electron gas .
therefore , the basic strategy which was intensively pursued already for more than decade was to incorporate the physics of on - site coulomb correlations in lda and to solve this problem using modern many - body techniques .
this way of thinking gave rise to such directions as lda@xmath15@xmath23 ( e.g. , refs .
, , and ) and lda@xmath15dmft ( dynamical mean - field theory , ref . ) . taking into account two above arguments , we believe that the most logical way to approach the problem of coulomb correlations in narrow - band compounds is to divide it in two part : + ( i ) mapping of conventional electronic structure calculations onto the multi - orbital hubbard model , and derivation of the parameters of this model from the first principles , for example starting from the simplest electronic structure in lda ; + ( ii ) solution of this multi - orbital hubbard model using modern many - body methods.@xcite in this paper we will discuss the first part of this project and show how results of conventional lda calculations for the @xmath3 bands can be mapped onto the multi - orbital hubbard model : @xmath24 where @xmath25 ( @xmath26 ) creates ( annihilates ) an electron in the wannier orbital @xmath27 of the site @xmath28 , and @xmath29 is a joint index , incorporating all remaining ( spin and orbital ) degrees of freedom . the matrix @xmath30 parameterizes the kinetic energy of electrons .
the matrix elements @xmath31 have the following meaning : the site - diagonal part ( @xmath28@xmath13@xmath32 ) describes the local level - splitting , caused by the crystal field and ( or ) the spin - orbit interaction , while the off - diagonal part ( @xmath28@xmath33@xmath32 ) stands for the transfer integrals ( or the transfer interactions ) .
@xmath34@xmath13@xmath35@xmath36@xmath37 are the matrix elements of _ screened _ coulomb interaction @xmath38@xmath36@xmath39 , which are supposed to be diagonal with respect to the site indices . in principle , the off - diagonal elements can be also included into the model .
however , we do not consider them in the present work . in sec .
[ sec : summary ] we will discuss several open questions related with the definition of the intersite coulomb interactions in lda . the first part of this paper will be devoted to derivation of the parameters of the kinetic energy .
then , we will explain how to construct the wannier functions ( wfs ) , which generate these parameters after applying to the ks hamiltonian in the real space .
the next part will be devoted to calculations of screened coulomb interactions , using the wf formalism .
in this section we will briefly review the main ideas of linear - muffin - tin - orbital ( lmto ) method , as they will be widely used in the subsequent sections for the construction of transfer interactions and the wannier orbitals .
the method is designed for the solution of ks equations in lda . for the details and recent developments ,
the reader is referred to the activity of the o. k. andersen group at the max - plank institute in stuttgart.@xcite majority of modern electronic structure methods use some basis .
the basis functions of the lmto method , @xmath40 ( the so - called muffin - tin orbitals mtos ) have many similarities with orthogonalized atomic orbitals .
as we will see below , the lmto method is very convenient for constructing the wfs , and for certain applications , the basis function of the lmto method , from the very beginning , can be chosen as a wannier function .
the conventional lmto approach employs the atomic - spheres - approximation ( asa ) , which assumes that the whole space of the crystal can be filled by overlapping atomic spheres ( fig .
[ fig.mto ] ) , so that the overlap between the spheres as well as the empty spaces , which are not encircled by any spheres , can be neglected .
the mtos are constructed from solutions of ks equations inside atomic spheres ( the partial waves ) , calculated at some energies @xmath41 ( typically , the center of gravity of the occupied band or of the entire band ) , @xmath42 , and their energy derivatives @xmath43 . in each atomic sphere , the ks potential is spherically averaged . therefore , the solutions are proportional to the angular harmonics , which are specified by the indices @xmath44@xmath45@xmath46 ( correspondingly , orbital and azimuthal quantum numbers ) . at the atomic sphere boundaries ,
@xmath47 and @xmath48 match continuously and differenciably onto certain envelop functions .
the latter are typically constructed from irregular solutions of laplace equation , which rapidly decay in the real space ( fig .
[ fig.mto ] ) .
it is easy to verify that the functions @xmath42 and @xmath43 obey the following `` lmto algebra '' : @xmath49 @xmath50 @xmath51 @xmath52 and @xmath53 where @xmath12@xmath13@xmath54 is the kohn - sham hamiltonian in the real space
. then , one possible choice of the muffin - tin orbitals is @xmath55 in this paper we use the shorthanded notations by andersen _
et al._. for instance , eq .
( [ eqn : lmtobasis1 ] ) should be read as follows : @xmath56 the first and second terms in the right - hand side of this equation are sometimes called , correspondingly , the `` head '' and the `` tail '' of mto .
the angular character of mto at the central site @xmath28 is specified by that of partial wave @xmath57 .
the matrix elements of @xmath12 in the mtos basis ( [ eqn : lmtobasis1 ] ) can be immediately derived by using the properties ( [ eqn : lmtoalgebra1])-([eqn : lmtoalgebra5 ] ) of @xmath57 and @xmath58 : @xmath59 where @xmath60 and @xmath61 are the diagonal matrices constructed from @xmath62 and @xmath63 , and @xmath64 is the shorthanded notation for the matrix @xmath65 .
the corresponding overlap matrix , @xmath66 , is @xmath67 since the second term in the right - hand side of @xmath68 is typically small , the basis functions ( [ eqn : lmtobasis1 ] ) are said to form a nearly orthogonal representation of the lmto method , and @xmath69 is the lmto hamiltonian in the second order of @xmath70@xmath36@xmath71 .
the basis ( [ eqn : lmtobasis1 ] ) can be orthonormalized numerically , by applying the transformation @xmath72 the corresponding lmto hamiltonian , @xmath73@xmath13@xmath74 , which is formally valid in all orders of @xmath70@xmath36@xmath71 , is given by @xmath75 this hamiltonian will be used as the starting point in the next section , for the definition of transfer interactions between certain wannier orbitals .
we start with the formal description of the downfolding method .
the construction of the wannier basis functions , underlying this approach , will be considered in sec .
[ sec : wannier ] , where we will use again all merits of the lmto method and show that proper wfs can be constructed by retaining the `` heads '' of mtos and attaching to them different `` tails '' . for periodic crystals ,
it is convenient to work in the reciprocal ( @xmath76 ) space . therefore , if it is not specified otherwise , we assume that the lmto hamiltonian is already constructed in the reciprocal space , after the fourier transformation of mtos : @xmath77 where @xmath78 is the number of sites . for all considered compounds ,
results of our asa - lmto calculations are in a good agreement the ones obtained using more accurate full - potential methods . in our definition of the crystal - field
splitting we go beyond the conventional asa and take into account nonsphericity of the electron - ion interactions ( see sec . [
sec : nsphcoulomb ] ) .
parameters of the kinetic energy are obtained using the downfolding method , starting from the electronic structure in lda . in order to describe properly the electronic structure of the tm oxides in the valent part of the spectrum using the lmto method , it is typically required several tens or even hundreds basis functions ( including the ones associated with empty spheres , which are added in order to improve the atomic spheres approximation for loosely packed atomic structures ) .
several examples of such bases will given in secs .
[ sec : tsrvo3]-[sec : tv2o3 ] . what we want to do next is to describe some part of this electronic structure by certain tight - binding ( tb ) hamiltonian @xmath79 , which , contrary @xmath73 , is formulated in the basis of a very limited number of orthogonal atomic - like orbitals .
for example , in order to reproduce the @xmath3 bands located near the fermi level , one would like to use only three @xmath3 orbitals centered at each tm site .
these orbitals have a meaning of wannier orbitals , which will be considered in sec .
[ sec : wannier ] .
we start with the identity by noting that any eigenstate of the lmto hamiltonian @xmath73 can be presented as the sum @xmath80@xmath13@xmath81@xmath15@xmath82 , where @xmath81 is expanded over the lmto basis function of the @xmath3-type , @xmath83 ( here , the character of the basis function is specified by its `` head '' ) , and @xmath82 is expanded over the rest of the basis functions @xmath84 .
then , the matrix equations for lmto eigenstates can be rearranged identically as @xmath85 by eliminating @xmath86 from eq .
( [ eqn : seceq2 ] ) one obtains the effective @xmath87-dependent hamiltonian in the basis of @xmath3-states @xmath88 and the `` overlap '' matrix @xmath89 satisfying the condition @xmath90@xmath13@xmath91 .
then , the required tb hamiltonian , @xmath79 , is obtained after the orthonormalization of the vectors @xmath92@xmath93@xmath94@xmath13@xmath95 and fixing the energy @xmath87 in the center of gravity of the @xmath3 band ( @xmath96 ) : @xmath97 typically , the downfolding is performed in the reciprocal space , and the parameter @xmath96 may also depend on @xmath76 .
the hamiltonian @xmath98 can be fourier transformed back to the real space : @xmath99 the site - diagonal part of @xmath100 shall describe the crystal - field ( cf ) splitting caused by the lattice distortion and associated with the transfer interactions between @xmath101- and @xmath102-orbitals , which are eliminated in the downfolding method , while the off - diagonal elements have a meaning of transfer interactions .
the first application of this approach has been considered in ref . .
in sec . [ sec : kineticapplications ] we will illustrate abilities of this method for several types of @xmath3 compounds .
the crystal - field splitting may have another origin , which is related with nonsphericity of the electron - ion interactions.@xcite this contribution will be considered in sec .
[ sec : nsphcoulomb ] .
in the previous section we have shown that there is a tb hamiltonian , @xmath79 , formulated in some basis of wannier orbitals @xmath103 .
it allows to generate the electronic structure of isolated @xmath3 bands , which is practically identical to the electronic structure obtained after the diagonalization of the total lmto hamiltonian @xmath73 . in this section
we will solve an inverse problem and construct the basis @xmath103 , which after applying to the original ks hamiltonian , generates the matrix @xmath79 : @xmath104 in a close analogy with the lmto method , we will first introduce the orbitals @xmath105 , which are related with @xmath106 by the orthonormalization transformation @xmath107 and search @xmath108 in the form : @xmath109 where @xmath110 is constructed entirely from the @xmath3-type solutions of ks equations inside atomic spheres and their energy - derivatives @xmath111 : i.e. both @xmath112 and @xmath113 belong to the tm sites , and ` t ' stands for the @xmath114-@xmath3 partial waves .
each @xmath115 is constructed from the rest of the partial waves @xmath116 .
then , @xmath110 and @xmath115 can be found from the following conditions : + 1 .
we request only the @xmath101-part of @xmath117 to contribute to the matrix elements of the ks hamiltonian , and search it in the form of mto : @xmath118 in this definition , @xmath119 is a function of ( yet unknown ) matrix @xmath120 , which will be found later .
the matrix elements of @xmath12 in the basis of these wannier orbitals are given by : @xmath121 2 .
the @xmath102-parts of the wf , @xmath122 , do not contribute to the matrix elements ( [ eqn : hwannierme2 ] ) .
they are introduced only in order to make the wfs ( [ eqn : wf2 ] ) orthogonal to the rest of the eigenstates of the hamiltonian @xmath73 .
therefore , we search @xmath123 in the form : @xmath124 where @xmath125 is obtained from the condition @xmath126@xmath13@xmath127 .
the coefficients @xmath128 are found from the orthogonality condition of @xmath117 to @xmath129 eigenstates @xmath130 of the original lmto hamiltonian @xmath73 : @xmath131 this allows to include the @xmath102-components of the wfs in a systematic way .
for example , by taking into consideration the @xmath21-partial waves inside oxygen spheres ( @xmath129@xmath13@xmath132 wavefunctions for cubic perovskites ) , the wfs can be orthogonalized to @xmath129@xmath13@xmath132 o(@xmath21 ) bands , etc . for a given @xmath120 ,
the problem is reduced to the solution of the system of linear equations ( [ eqn : orthogonality ] ) . since @xmath133 contributes to eq .
( [ eqn : orthogonality ] ) , the coefficients @xmath134 will also depend on @xmath120 .
therefore , the total wf is an implicit function of the matrix @xmath120 : @xmath117@xmath45@xmath135 .
the last step is the orthonormalization ( [ eqn : wf1 ] ) , which after substitution into eq .
( [ eqn : hwannierme1 ] ) yields the following equation for the matrix @xmath120 : @xmath136 this equation is solved iteratively with respect to @xmath120 .
the choice of the wfs as well as their extension in the real space is not uniquely defined . for
many practical applications one would like to have `` maximally localized '' orbitals,@xcite although in the context of the wfs , the term `` maximally localized '' itself bears certain arbitrariness and is merely a mathematical construction , because depending on the considered physical property one can introduce different criteria of the `` maximal localization '' .
although we do not explicitly employ here any procedure which would pick up the `` most localized '' representation for the wfs , our method well suits this general strategy and the obtained wfs are expected to be well localized around the central tm sites .
there are several ways of controlling the spacial extension of the wfs in the lmto method .
+ 1 . by using different envelop functions one can , in principle ,
change the spacial extension of mtos ( fig .
[ fig.mto ] ) , which controls the decay of the original lmto hamiltonian in the real space .
for example , instead of irregular solutions of laplace equation , one can use hankel functions of the complex argument .
however , any choice should satisfy certain criteria of the completeness of the basis set . from this point of view ,
the use of the hankel functions is not well justified as it typically deteriorates the accuracy of lmto calculations .
therefore , in the present work we leave this problem as it is and fix the lmto basis set .
once the lmto basis is fixed , the relative weight of the tm @xmath4-states and other atomic states which contribute to the @xmath3 band can not be changed ( see fig .
[ fig.dossummary ] ) .
for example , the contribution of the oxygen @xmath21-states can not be replaced by muffin - tin orbitals centered at the tm sites and vise versa .
the same proportion of atomic orbitals should be preserved in the wfs , constructed for this @xmath3 band .
then , the only parameter which can be controlled is how many wfs , centered at different sites of the lattice , contribute to the density of @xmath4-states at the given tm site .
then , the definition `` localized orbital '' mean that it is mainly centered around given tm site .
conversely , the `` delocalized orbital '' may have a long tail spreading over other tm sites .
then , it is easy to see that our procedure corresponds to the former choice .
indeed , in the first order of @xmath137@xmath36@xmath138 and neglecting for a while the nonorthogonality to the rest of the electronic states , the norm of the wf can be obtained from eq .
( [ eqn : wft ] ) as @xmath139@xmath13@xmath140 , meaning that the wf is fully localized at the central tm site .
then , it holds @xmath120@xmath13@xmath79 , which is valid in the second order of @xmath137@xmath36@xmath138.@xcite therefore , the leading correction to the above approximation , which define the actual weight of the wf at the neighboring tm sites is controlled by the parameters of the kinetic energy @xmath79 , and is of the order of @xmath141 . as we will see below ,
the latter is small .
the conclusion is rather generic and well anticipated for the strongly - correlated systems for which the kinetic - energy terms is generally small .
the angular character of the wf at the central tm site should be consistent with the one extracted from the local density of states in the region of @xmath3 bands ( in the other words , the local density of states at the tm sites should be well represented by atomic orbitals @xmath83 used in the downfolding method ) .
therefore , we choose @xmath83 as the set of atomic orbitals which mainly contribute to the local density of states in the region of @xmath3 bands .
for these purposes , at each tm site we calculate the density matrix in the basis of five @xmath4 orbitals @xmath142 : @xmath143 and sum up the contributions of all @xmath3 bands ( here , @xmath144 is an joint index , which incorporates the band index and the coordinates of the @xmath76-point in the first brillouin zone ) .
then , we diagonalize @xmath145 , and assign three most populated orbitals , obtained after the diagonalization to @xmath83 .
the contribution of coulomb interactions to the crystal - field splitting is a tricky issue . despite an apparent simplicity of the problem , one should clearly distinguish different contribution and not to include them twice , in the kinetic and coulomb parts of the model hamiltonian ( [ eqn : hmanybody ] ) .
the use of full - potential techniques does not automatically guarantee the right answer .
however , the atomic - spheres - approximation , which typically supplements the lmto method , will also require additional corrections for the crystal - field splitting . in this section
we would like to make two comments on this problem .
the nonsphericity of _ on - site _ coulomb interactions is already included in the second part of the model hamiltonian ( [ eqn : hmanybody ] ) .
the problem will be discussed in details in sec .
[ sec : screenedu ] .
therefore , in order to avoid the double counting , the corresponding contribution to the kinetic - energy part should be subtracted . from this point of view the use of the spherically averaged ks potential in asa
is well justified .
the same is true for the intersite coulomb interactions , if they are explicitly included to the model hamiltonian ( [ eqn : hmanybody ] ) .
all remaining interactions should generally contribute to the crystal - field splitting .
in asa , the proper correction at the site @xmath28 can be found by considering the matrix elements of the coulomb potential produced by all other atomic spheres ( or ions ) at the site @xmath28 , @xmath146@xmath45@xmath147 is the wf centered at the site @xmath28 , and @xmath148 is the total charge associated with the sphere @xmath32 : namely , the nuclear charge minus the electronic charge encircled by the atomic sphere . the nonspherical part of this integral can be easily calculated in the real space by using the multipole expansion for @xmath149 . in all forthcoming discussions , unless it is specified otherwise , the matrix elements of the crystal - field splitting will incorporate the correction @xmath150 associated with nonsphericity of the electron - ion interactions .
srvo@xmath0 is a rare example of perovskite compounds , which crystallizes in the ideal cubic structure .
it attracted a considerable attention in the connection with the bandwidth control of the metal - insulator transition.@xcite for the cubic compounds , the separation of the lmto basis functions into @xmath151 and @xmath84 used in the downfolding method is rather straightforward : three @xmath3 orbitals centered at each v site form the subspace of @xmath83 orbitals , and the rest of the basis functions are associated with @xmath84 .
parameters of lmto calculations for srvo@xmath0 are given in table [ tab.srvo3lmtoparameters ] .
.[tab.srvo3lmtoparameters ] atomic positions ( in units of cubic lattice parameter @xmath152@xmath13@xmath153 ) , atomic radii ( in ) and basis functions included in lmto calculations for cubic srvo@xmath0 . [ cols="^,^,^,^,^ " , ] after the fourier transformation to the real space , we obtain the following parameters of on - site and nn interactions : @xmath23@xmath13@xmath154 ev and @xmath155@xmath13@xmath156 ev .
this value of @xmath23 appears to be in a reasonable agreement with the final @xmath23@xmath13@xmath157 ev , extracted from rpa ( fig .
[ fig.srvo3dosurpa ] ) .
however , c - lda employs an additional atomic - spheres approximation .
therefore , for a proper comparison with rpa , one should use the same level of approximation and suppress all nonspherical interactions in the matrix of coulomb interactions @xmath158 , which is used as the starting point in rpa . in this approximation , and considering only the o(@xmath21)@xmath93v(@xmath22 ) transitions in the polarization function , we obtain @xmath23@xmath13@xmath159 ev , which is close to the c - lda value obtained using the method proposed in ref . .
the small difference is caused by different approximations used for treating the intersite coulomb interactions , which were neglected in rpa and taken into account in c - lda .
for comparison , the total value of @xmath23 obtained in rpa after neglecting the nonsphericity effects is only @xmath160 ev . in summarizing this section , there is a reasonable agreement between results of rpa calculations and the c - lda approach proposed in ref . .
however , the agreement is somewhat fortuitous because this c - lda takes into account only one part of the total screening , corresponding to the o(@xmath21)@xmath93tm(@xmath22 ) transitions in the polarization function .
the error caused by this approximation is partially compensated by the atomic - spheres approximation supplementing the c - lda scheme .
the c - lda calculations give some idea about the effect of hybridization on the screening of intersite coulomb interactions .
the screening appears to be very efficient .
so , by taking into account only the o(@xmath21)@xmath93v(@xmath22 ) transitions , the nn interaction is reduced from 1.2 to 0.3 ev .
we expect this value to be further reduced by including other types of transitions in the polarization function .
thus , for the considered compounds , the effective coulomb interaction between different tm sites seems to be small and can be neglected . in this section
we discuss the effects of electron / hole doping on the static coulomb interactions in srvo@xmath0 using the rigid - band approximation .
results of such c - lda+rpa calculations are shown in fig .
[ fig.srvo3rpadoping ] versus the total number of electrons in the tm(@xmath3 ) band , @xmath161 .
we monitor the behavior of three kanamori parameters:@xcite @xmath162 , @xmath163 , and @xmath164 .
the coulomb interactions reveal a monotonic behavior as the function of doping .
the screening is the most efficient when the @xmath3 band is empty ( @xmath161@xmath13@xmath127 ) .
the situation corresponds to srtio@xmath0 . in this case all o(@xmath21)@xmath93tm(@xmath3 ) transitions contribute to the screening in rpa ( see fig .
[ fig.srvo3dosurpa ] ) .
this channel of screening vanishes when the @xmath3 band becomes occupied ( @xmath161@xmath13@xmath165 ) .
then , the only possible screening is associated with the o(@xmath21)@xmath93tm(@xmath22 ) transitions in the polarization function , and the effective coulomb interaction becomes large .
the screening of off - diagonal matrix element @xmath164 practically does not depend on doping .
therefore , the well know kanamori rule , @xmath162@xmath13@xmath163@xmath15@xmath166 , which was originally established for atoms , works well in the cubic compounds , even after the screening of @xmath3 interactions by other electrons .
the present result also supports an old empirical rule suggesting that only the coulomb integral @xmath23 is sensitive to the crystal environment in solids .
the nonspherical interactions , which are responsible for hund s first and second rules , appears to be much closer to their atomic values.@xcite we have shown that the change of hybridization plays a very important role and strongly reduces the static value of @xmath23 .
however , this is only one part of the story because the same effect implies the strong frequency - dependence of the effective interaction , as it immediately follows from the kramers - kronig transformation in rpa:@xcite @xmath167 indeed , the difference @xmath168@xmath36@xmath169 $ ] at @xmath87@xmath13@xmath127 , which for the diagonal matrix elements of srvo@xmath0 is about 8.7 ev , should be related with the existence of the finite spectral weight of @xmath170 at finite @xmath87 .
these dependencies are shown in fig .
[ fig.srvo3wsomega ] for srvo@xmath0 .
the high - frequency part of @xmath171 can also contribute to the low - energy part of the spectrum through the self - energy effect.@xcite the latter can be evaluated in the gw approximation,@xcite where the self - energy is given by the convolution of @xmath172 with the one - particle green function @xmath173 for the @xmath3 band : @xmath174 for srvo@xmath0 , the diagonal matrix element of @xmath175 is also shown in fig .
[ fig.srvo3wsomega ] .
the low - frequency part of @xmath176 is small and can be neglected , while @xmath177 mainly contributes to the renormalization factor : @xmath178^{-1}.\ ] ] the latter is estimated as @xmath179 , for the diagonal matrix elements . in sec .
[ sec : tytio3 ] we already pointed out that the degree of hybridization between atomic tm(@xmath3 ) and o(@xmath21 ) states can differ substantially for different tm oxides .
a typical example is two isoelectronic perovskites : srvo@xmath0 and ytio@xmath0 .
the tm(@xmath3)-o(@xmath21 ) hybridization is stronger in srvo@xmath0 , because of two reasons : + ( i ) a direct proximity of o(@xmath21 ) and v(@xmath3 ) bands in srvo@xmath0 , which is expected for tetra - valent compounds ; + ( ii ) a strong orthorhombic distortion observed in ytio@xmath0 , which generally deteriorates the ti(@xmath3)-o(@xmath21 ) hybridization .
therefore , it is reasonable to expect a very different screening of on - site coulomb interactions in these two compounds .
this idea is nicely supported by results of rpa calculations shown in fig .
[ fig.wdistortion ] .
the static @xmath23 is larger ytio@xmath0 .
for example , the diagonal matrix element of @xmath171 is about 3.4 ev , against 2.5 ev in srvo@xmath0 .
this is despite the fact that c - lda has an opposite tendency .
generally , the value of @xmath180 in c - lda is expected to be larger for srvo@xmath0 rather than for ytio@xmath0 , due to different shape of the atomic @xmath114-wavefunctions in the three- and tetra - valent compounds.@xcite so , c - lda yields @xmath180@xmath13 8.9 and 10.1 ev , correspondingly for ytio@xmath0 and srvo@xmath0 .
these parameters have been used as the starting point in rpa , which results in an opposite trend for the static @xmath23 .
therefore , the rpa screening is more efficient in srvo@xmath0 , which is consistent with stronger tm(@xmath3)-o(@xmath21 ) hybridization in this compound .
on the other hand , the frequency - dependence of @xmath171 is weaker in ytio@xmath0 , as it immediately follows from the kramers - kronig transformation ( [ eqn : kramerskronig ] ) . finally , because of different hybridization of the @xmath3 orbitals in ytio@xmath0 , the diagonal matrix elements of @xmath171 are also different ( see fig .
[ fig.wdistortion ] ) . in this case
, there is some deviation from the kanamori rules .
the difference is small ( about 0.07 ev for diagonal matrix elements of @xmath171 at @xmath87@xmath13@xmath127 ) .
however , it may play some role in more delicate applications , such as the orbital magnetism in solids , for example.@xcite in sec .
[ sec : tv2o3 ] we introduced two possible models for the kinetic - energy part of v@xmath1o@xmath0 : the `` five - orbital model '' treats all v(@xmath114 ) bands on an equal footing , while the `` three - orbital model '' is limited by twelve v(@xmath3 ) bands , close to the fermi level .
we argued that parameters of the kinetic energy can be different for these two models .
the same is true for the effective coulomb interactions .
the rpa provides a very transparent explanation for this difference , which is based on the following arguments . recall that rpa incorporates the screening of on - site coulomb interactions caused by relaxation of the wavefunctions
this relaxation is treated analytically , using regular perturbation theory for the wavefunctions , which results in eq .
( [ eqn : polarizationpt ] ) for the polarization function.@xcite since the v(@xmath22 ) band is eliminated in the three - orbital model , the effective @xmath171 should include the screening caused by that change of the wavefunctions , which is formulated in terms of transitions between v(@xmath3 ) and v(@xmath22 ) bands in the perturbation - theory expansion .
in the five - orbital model , the v(@xmath22 ) band is included explicitly .
therefore , the relaxation caused by possible interactions between v(@xmath3 ) and v(@xmath22 ) bands will be automatically taken into account in the process of solution of the five - orbital model , and we should get rid of this parasitic screening at the stage the construction of the model hamiltonian . thus , the tree - orbital model will include an additional screening of on - site interactions , caused by the v(@xmath3)@xmath93v(@xmath22 ) transitions in the polarization function , which does not appear in the five - orbital model .
this screening appears to be rather efficient ( unlike in cubic perovskites considered in sec [ sec : rparesultssrvo3 ] ) .
so , the diagonal matrix element of static @xmath3 interactions is about 3.2 and 3.9 ev , for the three- and five - orbital model , respectively ( fig .
[ fig.wv2o3 ] ) . on the other hand , according to the kramers - kronig transformation ( [ eqn : kramerskronig ] ) , the frequency - dependence of the effective interaction is more important in the three - orbital model .
because of different hybridization of @xmath3 and @xmath181 states in the five - orbital model , the @xmath181 interactions are screened more efficiently ( the static interaction between @xmath181 orbitals is about 3.7 ev ) .
the ultimate goal of this work was to make a bridge between first - principle electronic structure calculations and the universe of hubbard parameters for strongly - correlated systems .
we have presented a comprehensive analysis of the problem , by starting with the brief description of the asa - lmto method for electronic structure calculations and ending up with realistic parameters of the kinetic - energy and the coulomb interactions for the series of tm oxides obtained on the basis of this lmto method .
a particular attention has been paid to the analysis of microscopic processes responsible for the screening of on - site coulomb interactions in oxide compounds .
our strategy consists of three steps : + ( i ) derivation of the kinetic - energy part of the hubbard model from the single - particle electronic structure in lda , using the downfolding method .
we have also considered corrections to the crystal - field splitting caused by nonsphericity of electron - ion interactions , beyond the conventional atomic - spheres - approximation .
+ ( ii ) construction of the wannier functions using results of the downfolding method . at this stage
we closely follow the idea of lmto method , and construct the wfs as the lmto basis functions , which after applying to the kohn - sham hamiltonian in the real space generate the matrix elements of the kinetic energy obtained in the downfolding method .
+ ( iii ) calculation of screened coulomb interactions using the concept of _ auxiliary _ wfs .
the latter are defined as the wannier orbitals for which the kinetic - energy part is set to be zero .
this construction allows to avoid the double counting of the kinetic - energy term , which is included explicitly in the hubbard model .
the screened coulomb interactions are calculated on the basis of a hybrid approach , combining the conventional constraint - lda with the random - phase approximation for treating the hybridization effects between atomic tm(@xmath114 ) and o(@xmath21 ) orbitals .
the latter play a very important role and yields a strong renormalization of the effective coulomb interaction for isolated @xmath3 band
. it also explains a strong material - dependence of this interaction , which is sensitive to the crystal environment in solids , the number of @xmath3 electrons , the valent state of the tm ions , etc .
taking into account a wide interest to the construction of effective lattice fermion models from the first principles , we would like to make a brief comparison with other works on a similar subject .
majority of methods start with the construction of the wfs , which are then used as the basis for calculations of the parameters of the kinetic energy and the coulomb interactions .
this is different from our approach , where we start with the kinetic - energy part , and only after that construct the wfs _ for a given set of parameters of the kinetic energy_. we believe that such an order is extremely important , as it allows us to control the contributions of the kinetic energy to the wfs and the coulomb interactions . among recent works ,
a considerable attention is paid to the method of marzari and vanderbilt , because it allows to control the spacial extension of the wfs .
very recently , schnell _ et al .
_ applied this method to calculations of the parameters of the hubbard hamiltonian for the series of @xmath114 transition metals.@xcite in each @xmath76-point of the brillouin zone , they constructed the wfs from all 16 bands of the lda hamiltonian , corresponding to the @xmath182 lmto basis . therefore , the total number of wannier orbitals was also 16 .
the coefficients of this expansion has been chosen so to minimize the square of the position operator , @xmath183@xmath13@xmath184 .
the obtained wfs were indeed well localized , and the parameter @xmath23 estimated for the @xmath114 bands was very close to the atomic value ( about 25 ev ) .
however , the corresponding wannier basis set is too large , that does not make a big difference from the original lmto basis set , for which one can also introduce a localized ( tight - binding ) representation.@xcite from the view point of numerical solution of the hubbard model using modern many - body techniques , it is still hardly feasible to work in the basis of 16 wannier orbitals per one tm site , while the simplest hartree - fock approximation is definitely not sufficient for the transition metals.@xcite it would be interesting to see how this method will work for the tm oxides , considered in the present work , where the physical basis set is limited by three wannier orbitals per one tm site .
for example , is it possible to construct the localized wannier orbitals for _ isolated @xmath3 bands _ in the tm oxides , which would be as good as the wannier orbitals derived for _ all bands _ ?
the problem is that when the number of bands decreases , the number of variational parameters for the optimization of the wannier orbitals will also decreases .
therefore , the wannier orbitals will generally become less localized , as it was demonstrated in our work for v@xmath1o@xmath0 .
also , when the problem is formulated in a reduced hilbert space of bands closest to the fermi level , it is very important to consider the screening of coulomb interactions by other bands , which comes from relaxation of the wavefunctions .
these effects are beyond the scopes of the work of schnell _ et al .
_ , who only considered the bare coulomb interactions . in order to construct the wfs ,
ku _ et al . _ employed the projector - operator scheme,@xcite which is basically the initial step of the method of marzari and vanderbilt , prior the optimization.@xcite in this scheme , each wannier orbital is generated by projecting a trial wavefunction , @xmath185 , onto a chosen subset of bands ( in our case , @xmath3 bands ) : @xmath186 since such orbitals are not orthonormal , the procedure is followed by the numerical orthonormalization , similar to the one described in sec .
[ sec : lmto ] .
these wfs have been used as the basis for the construction of the low - energy hamiltonian for the series of cuprates , like la@xmath187ba@xmath1cu@xmath1o@xmath188 , where the wannier basis consisted of a single orbital centered around each cu site .
a weak point of this approach is that it is difficult to assess the spacial extension of the wf , which crucially depends on the trial wavefunction , and can be affected by the orthonormalization . in some sense , the result strongly depends on authors intuition on how they choose the trial wavefunction . for example , the _ bare _ coulomb interaction obtained in ref . in the basis of
their wfs was only 7.5 ev , which is much smaller than the atomic coulomb integral for the cu(@xmath114 ) orbitals .
this means that the wfs are not well localized .
it is not clear at present , whether this is a result of the bad choice of the trial wavefunction , or there is a more fundamental problem related with the fact that a more compact representation for the wfs simply may not exist in this case .
note , that apart from the berry phase , there is no further parameters available for the optimization of the wfs in the single - orbital case .
the delocalization of the wfs gives rise to appreciable direct exchange interactions operating between different cu sites.@xcite this result , however , rises additional questions .
note , that the kinetic part of the hubbard model and the wfs are evaluated in lda , where the exchange - correlation potential is set to be _
local_. in principle , non - local effects can be already incorporated in lda , through the renormalization of parameters of the kinetic energy and the local interactions.@xcite therefore , it is not clear whether the nonlocal exchange interactions should be regarded as independent parameters of the hubbard hamiltonian or not .
finally , ku _ et al .
_ calculated only _ bare _ coulomb interactions .
they did not consider the screening of these interactions caused by relaxation effects , which are extremely important .
anisimov _ et .
al . _ employed a similar approach for the analysis of spectroscopic properties of tm oxides.@xcite they extracted only the kinetic - energy part of the hubbard hamiltonian , using the wfs constructed in the lmto basis , and treated the coulomb interaction @xmath23 as a parameter .
a completely different strategy has been proposed by andersen _ et .
_ , on the basis of their order-@xmath78 muffin - tin orbital ( nmto ) method , which is an extension of the lmto method.@xcite from the very beginning , they construct the nmto basis functions in certain energy interval as the wfs of the original ks hamiltonian .
pavarini _ et al .
_ applied this method to the series of @xmath189 perovskites.@xcite at present , it is not clear how the nonuniqueness of the wfs is reflected in the construction of the nmto basis set .
obviously , such a basis set is also not unique , and there is some freedom left for the localization of the wannier orbitals , which does not seem to be well controlled .
generally , our transfer integrals for the @xmath189 perovskites seems to be more localized and our crystal - field splitting is smaller .
for example , had we relaxed the constraint condition for the construction of the `` heads '' of the wfs , based on the diagonalization of the density matrix ( [ eqn : densitymatrix ] ) , our conclusion would have been also different : the crystal - field splitting would increase , but the transfer integrals would become less localized .
pavarini _ et al .
_ did not calculate the coulomb interactions .
instead , they used @xmath23@xmath190@xmath191 ev as a parameter , with the reference to the photoemission data.@xcite however , the photoemission data are typically interpreted in the cluster model , which treats the o(@xmath21 ) band _
explicitly_. for the isolated @xmath3 band , the effective interaction should include an additional renormalization coming from the relaxation of the o(@xmath21 ) band , which is _ eliminated _ in the @xmath3-model .
therefore , the value of the effective @xmath23 should be smaller .
finally , due to unknown for us reason , there is a substantial difference of the parameters of @xmath3 bandwidth ( @xmath192 ) between our work and ref . , even for cubic srvo@xmath0 .
the parameters reported by pavarini _ et al .
_ are generally overestimated by about 30%.@xcite our conclusions about magnetic properties of ytio@xmath0 and latio@xmath0 are also different.@xcite details will be presented in a separate paper .
i thank professor masatoshi imada for valuable suggestions , especially for drawing my attention to the problem of wannier functions , nonsphericity of electron - ion interactions and its effect on the crystal - field splitting,@xcite and the random - phase approximation for calculating the effective interaction parameters in the hubbard model.@xcite
in this appendix we consider the screening of on - site coulomb interactions in rpa for an @xmath193-orbital system .
all orbitals are supposed to be equivalent . for simplicity
we neglect small nonsphericity of bare coulomb interactions .
then , the nonvanishing matrix elements of the coulomb interactions are @xmath194 , where @xmath29 ( @xmath195)@xmath13@xmath196 .
they can be presented in the form @xmath158@xmath13@xmath197 , where @xmath198 is the @xmath193@xmath199@xmath193 matrix , consisting of only the units : @xmath200 the part of the polarization polarization matrix ( [ eqn : polarizationpt ] ) , which can interact with the matrix @xmath158 , is assumed to be diagonal : @xmath201 .
the assumption is justified for cubic perovskites , where different @xmath3 orbitals belong to different bands . for other compounds
it can be regarded as an approximation , which does not change our qualitative conclusion .
hence , for the screened coulomb interaction ( [ eqn : urpa ] ) we have : @xmath202^{-1}\hat{u } = \sum_{n=0}^\infty ( up)^n \hat{i}^n u \hat{i}. \label{eqn : appendix1}\ ] ] since @xmath203@xmath13@xmath204 , eq . ( [ eqn : appendix1 ] ) can be converted to @xmath205 this means that in the multi - orbital systems , the renormalization of the coulomb repulsion is more efficient as it is controlled by the quantity @xmath206 , where the prefactor @xmath193 stands for the number of orbitals .
v. i. anisimov , a. i. poteryaev , m. a. korotin , a. o. anokhin , and g. kotliar , j. phys
. condens .
* 9 * , 7359 ( 1997 ) ; s. y. savrasov , g. kotliar , and e. abrahams , nature * 410 * , 793 ( 2001 ) ; k. held , g. keller , v. eyert , d. vollhardt , and v. i. anisimov , phys .
lett . * 86 * , 5345 ( 2001 ) ; a. i. lichtenstein , m. i. katsnelson , and g. kotliar , phys .
lett . * 87 * , 067205 ( 2001 ) .
o. k. andersen , phys .
b * 12 * , 3060 ( 1975 ) ; o. gunnarsson , o. jepsen , and o. k. andersen , _ ibid .
_ * 27 * , 7144 ( 1983 ) ; o. k. andersen and o. jepsen , phys .
lett . * 53 * , 2571 ( 1984 ) ; o. k. andersen , in _ the electronic structure of complex systems _ , ed . by p.
phariseau and w. m. temmerman ( plenum publishing corporation , 1984 ) ; h. l. skriver , _ the lmto method _ , ( springer - verlag , berlin , 1984 ) ; o. k. andersen , z. pawlowska , and o. jepsen , phys . rev . b * 34 * , 5253 ( 1986 ) ; o. k. andersen , o. jepsen , and m. sob , in _ electronic band structure and its applications _
, ed . by m.
yussouff ( springer - verlag , berlin , 1986 ) ; homepage : http://www.fkf.mpg.de / andersen/. for example , one can construct the wfs simultaneously for the @xmath3 and o(@xmath21 ) bands of srvo@xmath0 , by starting with the v(@xmath114-@xmath3 ) and o(@xmath21 ) orbitals ( which form the @xmath101-part of the wfs ) and including the v(@xmath114-@xmath22 ) and sr(@xmath210 ) partial waves into the @xmath102-part , in order to enforce the orthogonality to other bands . in this case
we do not need to worry about extension of the wfs centered at the v sites onto the oxygen sites because the o(@xmath21 ) states are explicitely included into the wannier basis .
hence , the wfs can be made more localized .
indeed , in this case we obtain @xmath183@xmath13 0.66 and 1.87 @xmath211 for the wannier orbitals centered at the v and o sites , respectively .
these values are already comparable with those obtained in ref . for mno , using the procedure of marzari and vanderbilt .
note that this splitting is different from the `` two - down , one - up '' scheme of the atomic @xmath3 levels ( @xmath36@xmath212 , @xmath36@xmath213 , and @xmath214 ev ) reported in ref . .
the difference is related with nonsphericity of the electron - ion interaction which was not considered in ref . .
the new scheme of the crystal - field splitting tends to stabilize a single @xmath3 orbital and thereby suppress the orbital fluctuations . on the hartree - fock level , the magnetic ground state of ytio@xmath0 and latio@xmath0 is expected to ferromagnetic and a - type antiferromagnetic , respectively , in agreement with ref . .
other details will be reported in a separate paper .
j. s. gardner , b. d. gaulin , s .- h .
lee , c. broholm , n. p. raju , and j. e. greedan , phys .
* 83 * , 211 ( 1999 ) ; c. h. booth , j. s. gardner , g. h. kwei , r. h. heffner , f. bridges , and m. a. subramanian , phys . rev .
b * 62 * , r755 ( 2000 ) ; a. keren and j. s. gardner , phys .
lett . * 87 * , 177201 ( 2001 ) . in the atomic limit ,
the @xmath191@xmath199@xmath191@xmath199@xmath191@xmath199@xmath191 matrix @xmath158 of the coulomb interactions between @xmath4 electrons is totally specified by three radial slater s integral : @xmath217 , @xmath218 , and @xmath219 .
then , we use the following definitions for the parameters @xmath180 and @xmath220 in terms of @xmath217 , @xmath218 , and @xmath219 : @xmath180@xmath13@xmath217 and @xmath220@xmath13@xmath221@xmath15@xmath222 . in lda@xmath15@xmath23
, one can use the ratio:@xcite @xmath223@xmath224@xmath225 , which holds in the atomic limit .
it allows to obtain all three slater s integrals , using the values of only two parameters , @xmath180 and @xmath220 , extracted from c - lda .
we also list here several useful identities which hold for coulomb matrix elements between @xmath3 orbitals : @xmath226@xmath13@xmath217@xmath15@xmath227@xmath15@xmath228 ( @xmath229@xmath15@xmath230 ) , @xmath231@xmath13@xmath217@xmath36@xmath232@xmath36@xmath233 , and @xmath234@xmath13@xmath235@xmath15@xmath236 .
i. schnell , g. czycholl , and r. c. albers , phys . rev . b * 65 * , 075103 ( 2002 ) ; i. schnell , phd thesis ( university of bremen , germany , 2002 ) ; i. schnell , g. czycholl , and r. c. albers , phys .
b * 68 * , 245102 ( 2002 ) .
v. i. anisimov , d. e. kondakov , a. v. kozhevnikov , i. a. nekrasov , z. v. pchelkina , j. w. allen , s .- k .
kim , p. metcalf , s. suga , a. sekiyama , g. keller , i. leonov , x. ren , and d. vollhardt , phys .
b * 71 * , 125119 ( 2005 ) .
e. pavarini , s. biermann , a. poteryaev , a. i. lichtenstein , a. georges , and o. k. andersen , phys .
92 * , 176403 ( 2004 ) ; e. pavarini , a. yamasaki , j. nuss , and o. k. andersen , cond - mat/0504034 . for srvo@xmath0
we have obtained @xmath237@xmath13@xmath238 ev , while pavarini _ et al .
_ reported @xmath237@xmath13@xmath239 ev.@xcite the electronic structure of cubic perovskites has been systematicaly studied by k. takegahara [ j. electron spectrosc .
relat . phenom .
* 66 * , 303 ( 1994 ) ] , using the augmented plane wave method within muffin - tin approximation . for srvo@xmath0
, he had @xmath237@xmath190@xmath240 ev .
this electronic structure has been used in dmft calculations by a. liebsch [ phys .
lett . * 90 * , 096401 ( 2003 ) ] .
there is another cubic perovskite , srtio@xmath0 , which has been intensively investigated by first - principle electronic structute calculations . for this material ,
the full - potential calculations yield @xmath237@xmath190@xmath157 ev [ e.g. , s. kimura _ et al .
_ , phys .
b * 51 * , 11049 ( 1995 ) ] , which is in fair agreement with our finding @xmath237@xmath13@xmath240 ev .
takegahara s value is @xmath237@xmath13@xmath157 ev .
the orthorhombic ytio@xmath0 has been studied in details by h. sawada and k. terakura [ phys .
b * 58 * , 6831 ( 1998 ) ] . they have obtained @xmath237@xmath190@xmath241 ev .
it is important to note that this value corresponds to the generalized gradient approximation ( gga ) , which additionally splits the @xmath3-bands because of strong nonsphericity of the exhcnage - correlation potential .
this nonsphericity should be suppressed in the process of construction of the hubbard hamiltonian ( see sec . [
sec : nsphcoulomb ] ) .
therefore , our value @xmath237@xmath190@xmath242 ev for ytio@xmath0 seems to be reasonable . on the other hand , pavarini _ et al . _
reported ( using the atomic - spheres approximation , similar to ours ) @xmath237@xmath13@xmath243 , which was even larger than the @xmath3-bandwidth of the considerably less distorted latio@xmath0 .
also , for latio@xmath0 we did not observe any overlap between ti(@xmath3 ) and la(@xmath244 ) bands , reported in ref . . | we propose a systematic procedure for constructing effective lattice fermion models for narrow - band compounds on the basis of first - principles electronic - structure calculations .
the method is illustrated for the series of transition - metal ( tm ) oxides : srvo@xmath0 , ytio@xmath0 , v@xmath1o@xmath0 , and y@xmath1mo@xmath1o@xmath2 , whose low - energy properties are linked exclusively to the electronic structure of an isolated @xmath3 band .
the method consists of three parts , starting from the electronic structure in the local - density approximation ( lda ) .
( i ) construction of the kinetic - energy hamiltonian using formal downfolding method .
it allows to describe the band structure close to the fermi level in terms of a limited number of ( unknown yet ) wannier functions ( wfs ) , and eliminate the rest of the basis states .
( ii ) solution of an inverse problem and construction of wfs for the given kinetic - energy hamiltonian . here
, we closely follow the construction of the basis functions in the liner - muffin - tin - orbital ( lmto ) method , and enforce the orthogonality of wfs to other band . in this approach
, one can easily control the contributions of the kinetic energy to the wfs .
( iii ) calculation of screened coulomb interactions in the basis of _ auxiliary _ wfs .
the latter are defined as the wfs for which the kinetic - energy term is set to be zero . meanwhile
, the hybridization between tm @xmath4 and other atomic states is well preserved by the orthogonality condition to other bands .
the use of auxiliary wfs is necessary in order to avoid the double counting of the kinetic - energy term , which is included explicitly in the model hamiltonian . in order to calculate the screened coulomb interactions we employed a hybrid approach .
first , we evaluate the screening caused by the change of occupation numbers and the relaxation of the lmto basis functions , using the conventional constraint - lda approach , where all matrix elements of hybridization connecting the tm @xmath4 orbitals and other orbitals are set to be zero .
then , we switch on the hybridization and evaluate the screening of on - site coulomb interactions associated with the change of this hybridization in the random - phase approximation .
the second channel of screening appears to be very important , and results in relatively small value of the effective coulomb interaction for isolated @xmath3 bands ( about 2 - 3 ev , depending on the material ) .
we discuss details of this screening and consider its band - filling dependence , frequency dependence , influence of the lattice distortion , proximity of other bands , as well as the effect of dimensionality of the model hamiltonian . |
high - energy proton - nucleus ( pa ) collisions allow us to explore the dense gluon system appearing at small values of the bjorken @xmath0 in the target nucleus .
such a dense gluon system is expected to possess a universal feature of _ parton saturation _ , characterized by the saturation momentum scale @xmath3 , and has been investigated with the color glass condensate ( cgc ) effective theory@xcite . in a heavy nucleus with atomic mass number @xmath4 ,
the saturation phenomenon will become relevant even at moderate @xmath0 because of its larger gluon density by a factor of target thickness @xmath5 .
very recently p+pb collisions at the center - of - mass energy @xmath6 tev have been delivered at the large hadron collider ( lhc ) , and new exciting data are being reported such as hadron multiplicity@xcite and ridge phenomenon@xcite . the observed hadron multiplicity and momentum spectrum
@xcite will constrain theoretical models . at this high energy ,
the relevant value of @xmath7 becomes so small that the parton saturation scale @xmath8 with @xmath9@xcite in the nuclear target will be comparable to or larger than the charm quark mass @xmath10 .
this suggests coherence effects even in charm quark production , and thus we can get useful information of the gluon saturation in the nuclear target by studying the energy and rapidity dependences of charm quark and quarkonium productions in pa collisions at relativistic heavy ion collider ( rhic ) and the lhc@xcite . in heavy ion collision experiments , heavy quark and quarkonium productions@xcite
are very valuable probes for quantifying properties of hot and dense matter or a quark - gluon plasma transiently created in the events .
quarkonium suppression@xcite and enhancement@xcite , and energy loss@xcite and collective flow of heavy flavor mesons@xcite have already been discussed extensively .
here one obviously needs to know the initial nuclear effects on their productions for quantitative understanding of hot - medium modifications . in this paper
we shall investigate phenomenological implications of the saturation effects on the quarkonium production at collider energies by exploiting the color evaporation model ( cem)@xcite .
systematic study of quarkonium production spectrum from cgc in pa collisions will quantitatively improve our understanding of gluon saturation effects on particle production in pa collisions . at the same time
, it will serve as a benchmark for assessing the hot medium effects on the quarkonium production in aa collisions . in the cgc framework ,
the quark - pair production cross section in pa collisions is obtained by blaizot , gelis and venugopalan @xcite , where pa is treated as a `` dilute - dense '' system and the cross section is evaluated at leading order in the coupling constant @xmath11 and the color charge density @xmath12 in the proton , but in full orders with respect to the color charge density @xmath13 in the nucleus . adopting this formula , we previously evaluated the heavy quark production cross - sections in high - energy pa collisions to reveal their general features@xcite . of particular importance is the @xmath0 dependence of the unintegrated gluon distribution ( ugd ) in the nuclear target .
we describe the @xmath0 dependence of ugd with the balitsky - kovchegov ( bk ) equation including the running coupling corrections ( rcbk equation)@xcite . the bk equation can be derived from a more general equation of functional renormalization group by taking the mean - field approximation .
the authors of @xcite analyzed the global data of deep inelastic scatterings ( dis ) of e+p at hera using the rcbk equation , to obtain a set of constrained ugd of the proton at small @xmath14 .
the resultant ugd has been applied to compute the particle production at the proton-(anti)proton colliders@xcite .
we use this constrained ugd in this work . in forward particle production ,
the momentum fraction @xmath15 of the gluons from the proton is not small , and the distribution may be better described with the ordinary collinear gluon distribution function .
accordingly , the expression of the cross section is obtained by taking the collinear limit @xmath16 for the gluon momentum from the proton in the hard matrix element .
( this kind of asymmetric treatment is well known for the hadron production from cgc@xcite . )
we will compare the quarkonium production in the collinear approximation on the proton side with the original one that involves the @xmath17 dependent ugd for the proton .
this paper is organized as follows : in sec . 2 we review the quark pair production formula in the cgc framework and explain its collinear limit on the proton side .
then we incorporate the @xmath0-dependence of ugd using the rcbk evolution equation . in sec . 3 , we compute the quarkonium production cross section working in the color evaporation model , and show the transverse momentum spectra and nuclear modification factor for j/@xmath1 at rhic and lhc energies , and those for @xmath18 at lhc energy .
rapidity and initial - saturation - scale dependences of the nuclear modification factor , and momentum broadening are also discussed .
is devoted to conclusion and outlook .
in this section we summarize analytic expressions for the quark - pair production cross - section , derived in ref .
we also present its limit when the transverse momentum of the gluon coming from the proton is small , in order to make use of conventional collinear factorization for the proton . for more details ,
see refs .
@xcite . in the cgc formalism
, the proton - nucleus collision is described as a collision of two sets of color sources representing the large @xmath0 degrees of freedom in the proton and the nucleus respectively .
when they collide , these color sources produce a time - dependent classical color field , and this color field can in turn produce quark - antiquark pairs .
both the classical color field and the quark - pair production amplitude can be calculated analytically when one of the projectiles is dilute and its density of color sources can be treated to the lowest order .
this is the approximation we make in the description of proton - nucleus collisions in this framework , the proton naturally being the dilute projectile . for definiteness ,
let us denote by @xmath12 and @xmath20 the densities of color sources in the proton and the nucleus respectively , and that the proton moves in the @xmath21 direction and the nucleus in the @xmath22 direction .
then , the pair production amplitude reads @xcite : @xmath23 + t_{g}(\k_{1\perp})[t^bu^{ba}(\x_\perp)]\bigg\ } v(\p ) \ ; , \label{eq : mf - final-1}\end{aligned}\ ] ] where we denote and @xmath24 of the produced particles have not been listed among the arguments of these objects in order to make the equations more compact . ] @xmath25 + 2q^+[(\q_\perp\!-\!\k_\perp\!-\!\k_{1\perp})^2+m^2 ] } \ ; , { \nonumber\\}&&t_{g}(\k_{1\perp})\equiv \frac{{\raise.15ex\hbox{$/$}\kern-.53em\hbox{$c$}}_{_{l}}(p+q,\k_{1\perp})}{(p+q)^2 } \ ; , \label{eq : tqqbar - tg}\end{aligned}\ ] ] with @xmath26 , and where @xmath27 is the well - known lipatov effective vertex ] ] .
the matrix @xmath28 is the path - ordered exponential of the gauge field generated by the charge density @xmath20 in the nucleus : @xmath29 \ , , \end{aligned}\ ] ] with @xmath30 the su(@xmath19 ) generators in the fundamental representation , and @xmath31 is the same but in the adjoint representation .
the two terms in the curly bracket of eq .
( [ eq : mf - final-1 ] ) may be interpreted as the processes where a gluon from the proton emits the pair either before or after the collision with the nucleus .
the eikonal phases @xmath32 and @xmath33 describe the multiple scatterings of a quark or a gluon in the target nucleus .
an important property of this amplitude is that the sum of the two terms in the bracket vanishes when @xmath34 in agreement with ward identities @xcite this property is essential in order to recover the limit of collinear factorization on the proton side , which should hold since there is only a single scattering on the proton .
the probability of pair production is then obtained by squaring the above amplitude , and by averaging the squared amplitude with the classical charge distributions of the proton and the nucleus , @xmath35 $ ] and @xmath36 $ ] .
the resulting expression generally involves the nuclear multi - parton correlators , e.g. , @xmath37 , where @xmath38 indicates that the average over @xmath20 is performed with the distribution @xmath36 $ ] evolved to the rapidity @xmath39 .
2- and 3-point correlators are also needed in the calculation of the pair production probability , but they can be obtained as special cases of the 4-point function thanks to the identity @xmath40 .
this formula also leads to a general sum rule for the fourier transforms of these correlators @xmath41 , @xmath42 and @xmath43 @xcite .
all these correlators can be evaluated in a closed form when the distribution of color sources , @xmath36 $ ] , is a gaussian regime at very high energy .
we therefore expect that this is a reasonable approximation for our work . ]
( see @xcite ) , but the 4-point correlator has a very complicated expression which is quite hard to evaluate numerically .
however , in the large @xmath19 limit , it simplifies into @xmath44 with @xmath45 as one can see , it is possible in this limit to write the 4-point ( and also the 3-point ) function in terms of the 2-point function only , which simplifies considerably all the numerical calculations .
together with the translational invariance in the transverse plane , this fact makes the relation between @xmath41 and @xmath42 trivial . by exploiting these relations between the 4- , 3- and 2-point correlators , the pair production probability at the impact parameter @xmath46 , in the large @xmath19 limit , can be written as @xcite @xmath47 where we denote @xmath48 , @xmath49 the dimension of the adjoint representation of @xmath50 , and @xmath51 .
the variables @xmath52 are the rapidities of the gluons that come from the proton and from the nucleus respectively . a shorthand notation for the squared matrix element
is introduced as , @xmath53 and @xmath54 ( see @xcite ) . however , in the large @xmath19 limit , this 4-point function is in fact given by ( see @xcite ) @xmath55 which allowed us to equate @xmath53 and @xmath54 in the squared amplitude and to perform directly the @xmath54 integration . ]
@xmath56 { \nonumber\\ } & & + { \rm tr}_{\rm d } \big[({\raise.1ex\hbox{$/$}\kern-.53em\hbox{$q$}}\!+\!m)t_{q\bar{q}}({\raise.1ex\hbox{$/$}\kern-.63em\hbox{$p$}}\!-\!m ) \gamma^0 t_{g}^{\dagger}\gamma^0 + { \rm h.c.}\big ] { \nonumber\\ } & & + { \rm tr}_{\rm d } \big[({\raise.1ex\hbox{$/$}\kern-.53em\hbox{$q$}}\!+\!m)t_{g}({\raise.1ex\hbox{$/$}\kern-.63em\hbox{$p$}}\!-\!m)\gamma^0 t_{g}^{\dagger}\gamma^0\big ] \ ; .
\label{eq : xi}\end{aligned}\ ] ] in eq .
( [ eq : average - number - ln ] ) , @xmath57 is the ugd in the proton , and @xmath58 is expressed in terms of the fourier transform of the 3-point nuclear correlator as ) , this object appears in differential form with respect to the transverse coordinate @xmath59 because we are considering here the pair production probability at a fixed impact parameter @xmath46 . when we integrate it over @xmath46 in order to obtain the cross - section
, we will have to integrate this correlator over the transverse area of the nucleus . ] : @xmath60 where @xmath61 is the fourier transform of @xmath62 .
the @xmath59 dependence is rather weak for a large nucleus and may be treated implicitly through the variations of saturation scale @xmath63 with @xmath59 . in the second line of eq .
( [ eq:3ptfn ] ) , we have ignored the @xmath59-dependence of @xmath64 when doing the @xmath59 integration because the proton radius is small compared with that of a heavy nucleus ( @xmath65 ) .
we thus have a compact expression for the quark production probability , but our formula involves a nuclear 3-point function @xmath66 , which violates the usual form for @xmath67-factorization even in the leading order approximation@xcite .
the pair cross - section in the minimum - bias pa collision is obtained by integrating eq .
( [ eq : average - number - ln ] ) over the impact parameter @xmath46 . dividing the cross - section with the total inelastic cross - section @xmath68 , which we estimate as @xmath69
, we have the average multiplicity per event : @xmath70 here we have introduced the 3-point function integrated over nuclear transverse area @xmath71 which is related to the ugd ( 2-point function ) of the nucleus by @xmath72 .
the proton ugd @xmath73 may be estimated by replacing the transverse area @xmath74 and the amplitude @xmath75 with those for the proton . when the momentum fraction @xmath15 probed in the proton
is not small ( e.g. , @xmath76 ) , and even more so in the forward rapidity region where @xmath77 , the typical transverse momentum of the gluons in the proton is much smaller than the transverse mass of the produced quark or antiquark , @xmath78 .
we can therefore neglect @xmath17 in the matrix element @xmath79 in eq .
( [ eq : xi ] ) , and take the collinear approximation on the proton side .
this limit is well defined thanks to the fact that the expression on the second line in the amplitude in eq .
( [ eq : mf - final-1 ] ) goes to zero as @xmath34 @xcite : @xmath80 thus , the amplitude squared @xmath79 is quadratic in @xmath17 when @xmath34 , which cancels the factor @xmath81 in the denominator of eq .
( [ eq : cross - section - ln ] ) .
note that the `` vector '' @xmath82 in this formula contains spinors and dirac matrices . in this approximation , we can write the integral in eq .
( [ eq : cross - section - ln ] ) as @xmath83 where it is now implicit that @xmath84 should not exceed the typical transverse momentum scale set by the produced final state . using @xmath85 and @xmath86
we obtain @xmath87 where we have now @xmath88 , and where we denote @xmath89 .
the squared matrix element @xmath90 in the collinear approximation can be obtained by expanding the amplitude in eq .
( [ eq : mf - final-1 ] ) to linear order in the transverse momentum @xmath17 : @xmath91 with @xmath92\ ; , \nonumber\\ \xi_{\rm coll}^{q\bar{q},g } & = & -\frac{16}{(p+q)^2(\a_\perp^2+m^2 ) } \left [ m^2+\frac{(p^+)^2+(q^+)^2}{(p^++q^+)^3}\a_\perp\cdot(p^+\q_\perp - q^+\p_\perp ) \right]\ ; , \nonumber\\ \xi_{\rm coll}^{g , g } & = & \frac{8}{(p+q)^4 } \left [ ( p+q)^2-\frac{2}{(p^++q^+)^2}(p^+\q_\perp - q^+\p_\perp)^2 \right]\ ; .\end{aligned}\ ] ] in these formulas , we denote @xmath93 , and the squared invariant mass of the pair , @xmath94 , is given by @xmath95-(\p_\perp+\q_\perp)^2\ ; .\ ] ] the dense nuclear distribution of gluons is encoded in the 3-point correlator @xmath96 , that appears in the pair cross - section ( [ eq : cross - section - ln ] ) or ( [ eq : cross - section - ln - coll ] ) . in the large @xmath19 limit , it can be expressed in terms of the 2-point correlation @xmath61 , i.e. the forward scattering amplitude of a dipole in the fundamental representation . in the quasi - classical mv model ,
@xmath97 is independent of the rapidity variable @xmath98 , and it only includes the effects of the multiple scatterings of the quark - antiquark pair passing though the target nucleus , in the eikonal approximation . at large @xmath99 , the effect of multiple scatterings becomes small and leading twist results are recovered .
the energy ( rapidity ) dependence of @xmath97 arises via quantum fluctuations .
generally , the evolution equation for the 2-point function involves an infinite hierarchy of multi - point correlators .
however , it is known that in the limit of a large number of colors @xmath19 and of a large nucleus the energy evolution of @xmath61 can be described by a closed mean - field equation known as the balitsky - kovchegov ( bk ) equation@xcite .
this equation is an integro - differential equation that reads @xmath100\ ; , \label{eq : kovchegov}\end{aligned}\ ] ] where @xmath101 and @xmath102 is the evolution kernel ( see below ) .
thus , with an appropriate initial condition at a certain @xmath103 , we can consistently treat the rapidity dependence of the cross - section by substituting into eq .
( [ eq:3ptfn ] ) the solution @xmath61 of the bk equation for @xmath104 .
it is also well known that the bk equation with a fixed coupling constant requires a very low value of @xmath105 in order for the evolution of the saturation scale to be compatible with what one infers from hera data , _
i.e. _ , @xmath106 with @xmath107 @xcite .
it was argued that the next - to - leading order corrections to the bk equation would give the correct evolution speed with more reasonable values of @xmath105 @xcite .
indeed , it has been demonstrated recently in refs .
@xcite that the bk equation including the running coupling corrections in the kernel in balitsky s prescription@xcite : @xmath108 \ ; , \end{aligned}\ ] ] makes the saturation scale behave compatible with hera data , and the @xmath0-evolution equation becomes now a very useful tool ( called rcbk equation ) for phenomenology .
global fitting of the compiled hera e+p data at @xmath109 was performed with the rcbk equation in @xcite .
following their approach , we choose the initial condition at @xmath110 as @xmath111 \ ; , \label{eq : initialcondition}\end{aligned}\ ] ] and the parameter values are listed in table [ tab : par]@xcite . as for the running coupling constant in the evolution kernel we adopt the following form in the coordinate space : @xmath112^{-1}\end{aligned}\ ] ] with @xmath113 .
a constant @xmath114 is introduced so as to freeze the coupling constant smoothly at @xmath115 . the non - gaussian value @xmath116 is preferred by the fitting .
it is argued in @xcite that the value @xmath116 suggests a possible importance of higher - order color correlations in the proton and is valid for moderate values of transverse momentum @xmath117 .
we also list the mclerran - venugopalan model @xmath118 for comparison .
.parameter values for the two - point correlator @xmath75 .
@xmath119 gev is fixed .
[ tab : par ] [ cols="^,^,^,^,^",options="header " , ] for @xmath120 , we extrapolate the function @xmath96 with the following phenomenological ansatz @xcite : @xmath121 in this formula , the power @xmath122 for the factor @xmath123 comes from the behavior at large @xmath0 of the gluon distributions , as inferred from sum rules .
note that this extrapolation implies that the saturation scale is frozen at large @xmath0 , which may lead to a harder @xmath53-spectrum for @xmath124 than expected , possibly overestimating the cronin peak .
the saturation scale for a heavy nucleus at moderate values of @xmath0 will be enhanced by a factor of the nuclear thickness @xmath125 : @xmath126 @xcite . as we consider only mean bias events in this work , we assume a simpler relation @xmath127 where @xmath4 is the mass number of the nucleus under consideration .
this relation is assumed to be valid at the @xmath128 where the initial condition for the rcbk equation is set .
the rcbk equation then controls how the saturation scale of the nucleus evolves to lower values of @xmath0 .
taking into account the uncertainty of the saturation scale for a heavy nucleus @xmath129 at @xmath103 , we will consider the values in the range @xmath130 at @xmath131 . in fig .
[ fig : updf ] , we show the profile of ugds , @xmath132 and @xmath133 , of the proton and the nucleus , respectively , at several values of @xmath134 , obtained by solving the rcbk equation with set g1118 in table [ tab : par ] .
we also show the result of the bk evolution with the fixed coupling constant @xmath135 , for comparison .
this small coupling is necessary to keep the evolution speed compatible with the empirical value in the bk equation .
we see that with increasing rapidity @xmath136 the peak position ( _ i.e. , _ the saturation scale ) drifts slightly faster in the rcbk evolution than that in the bk evolution with @xmath135 , and that the former yields a steeper @xmath67-slope than the latter .
note that the ugd should behave as @xmath137 at large @xmath67 as is known in the lo double log approximation for bfkl or dglap equation , while the bfkl ( equivalently bk in the linear regime ) evolution gives harder @xmath67 spectrum .
a dip structure around @xmath138 gev at @xmath103 is caused by the parameter @xmath116 , but is soon smeared out in the evolution . in the nucleus case , we assumed the initial saturation scale @xmath139 at @xmath140 .
the ugd is more suppressed in low @xmath67 region and the peak position locates at larger @xmath67 than in the proton case , reflecting the stronger multiple scatterings in the nuclear target .
we consider here the @xmath7 coverage of the charm pair production in the plane of the rapidity @xmath136 and the transverse momentum @xmath141 of the pair at collision energies @xmath142=200 gev and 5.02 tev in fig .
[ fig : x2-coverage ] . here
we fix the pair s invariant mass @xmath143 gev , and draw the curves determined by @xmath144 , on which either @xmath15 or @xmath145 is constant .
the kinematically disallowed region where @xmath146 is indicated by the shaded area .
we see that , at the rhic energy , j/@xmath1 is produced from the gluons of moderate @xmath147 at mid - rapidities , while at forward rapidities @xmath148 the process gets sensitivity to the gluons at small @xmath149 . at the lhc energy , on the other hand , j/@xmath1 production is already sensitive to the small @xmath145 gluon even at mid - rapidity , and at forward rapidity it probes @xmath145 as low as @xmath150 . in the next section
, we study the quarkonium production in cem applied for the heavy - quark production cross - section ( [ eq : cross - section - ln ] ) or ( [ eq : cross - section - ln - coll ] ) with the @xmath0-evolved ugd @xmath151 .
cem has been successful in describing the j/@xmath1 production in high - energy proton - proton ( pp ) collisions .
we estimate the j/@xmath1 production from the quark - pair production cross section within cem@xcite : @xmath152 where @xmath10 ( @xmath153 ) is the charm quark ( @xmath154-meson ) mass .
a phenomenological constant @xmath155 represents the nonperturbative transition rate for the charm pairs , produced in the invariant mass range from @xmath156 to the threshold @xmath157 , to bound into a quarkonium .
its empirical value is around @xmath1580.05@xcite .
use of cem for pa collisions assumes that the bound state formation occurs outside the target nucleus .
we fix the threshold with @xmath153 ( @xmath159 ) = 1.864 ( 5.280 ) gev for j/@xmath1 ( @xmath18 ) .
a remark is here in order . in the pair - production cross section ( [ eq :
cross - section - ln ] ) the inelastic cross section estimated as @xmath160 in the denominator effectively cancels out with the same factor in @xmath161 , and that the cross section is now proportional to the effective transverse area @xmath162 of the proton appearing in @xmath73 . in the following calculations ,
we choose the proton size @xmath163 fm and the j/@xmath1 formation fraction @xmath164 as representative values .
one should keep in mind that the absolute normalization of the cross section depends on these parameters .
we also cancels @xmath105 in front of the cross section by @xmath105 appearing in the denominator in @xmath165 and @xmath73 . in the case of collinear approximation on the proton side , we set @xmath166 in this paper .
we first show in fig .
[ fig : jpsi_pt_rhic_pp ] the transverse momentum spectrum of the produced j/@xmath1 in pp collisions ( see text ) . ] at @xmath167 gev , using the ugd set g1118 given in table [ tab : par ] .
the upper ( lower ) curve of each band indicates the result with charm quark mass @xmath168 ( 1.5 ) gev . in the collinear approximation on the larger-@xmath15 side ,
we adopt cteq6lo parametrization@xcite , and the band in fig .
[ fig : jpsi_pt_rhic_pp ] includes the change of the factorization scale from @xmath169 to @xmath170 with @xmath171 , where @xmath172 is the pair s invariant mass . as mentioned above , the quarkonium production at mid - rapidity @xmath173
is largely determined by the gluon distributions at moderate @xmath174 .
then , we notice a difficulty with set g1118 : the peculiar dip structure of g1118 seen in fig .
[ fig : bksol ] remains in the j/@xmath1 spectrum as a similar dip around @xmath175 gev , which must be an artifact of this initial condition .
in contrast , we do nt see such a structure with the mv initial condition . at forward - rapidity @xmath176 ,
the dip is smeared to be less noticeable by the imbalance between @xmath15 and @xmath145 and by the @xmath145 evolution of the ugd . as a whole ,
the @xmath141 spectrum obtained with set g1118 is closer to the observed data @xcite than with set mv . in this pp case
, the collinear approximation on the large-@xmath15 side does not improve the description of the data .
the @xmath67 kick from only the one of the protons can not give enough @xmath141 for the pair . in fig .
[ fig : jpsi_pt ] shown is the transverse momentum spectrum of the j/@xmath1 in pa collisions in our model .
we set the initial saturation scale of the ugd for the heavy nucleus as @xmath177 at @xmath103 .
the upper ( lower ) curve of each band indicates the result with @xmath168 ( 1.5 ) gev .
we overlay d - au data observed by phenix at @xmath167 gev@xcite , presuming here that the difference between pa and da results only in normalization difference of order o(1 ) .
we find that @xmath141-dependence of j/@xmath1 production is better described with set g1118 by the multiple scattering effects in the nuclear ugd . ] than that with set mv . indeed , here
the collinear approximation on the proton side apparently gives a better description of the data both at mid- and forward - rapidity regions .
however , at forward rapidities , where we approach the small-@xmath145 region and the kinematical boundary for @xmath15 at the same time ( see fig .
[ fig : x2-coverage ] ) , we expect a nontrivial interplay between large @xmath15 and small @xmath145 . besides the saturation dynamics of @xmath145 gluons , one may need to consider other physics such as energy loss of large-@xmath15 gluons in the heavy target@xcite in order to understand the @xmath141 spectrum of j/@xmath1 in the very forward region .
these effects are not included in our present treatment .
we notice in fig .
[ fig : jpsi_pt ] ( b ) that the j/@xmath1 production is more suppressed nearly by one order of magnitude in the collinear approximation than those in the full calculation .
this is caused by a difference in the large @xmath15 behavior of gluon distributions on the proton side .
as @xmath178 , the cteq gluon distribution decreases much more rapidly than our model ugd @xmath73 , which is assumed as @xmath179 .
furthermore , in the collinear approximation , the pair s @xmath141 is entirely provided from the nucleus side , @xmath180 , and ugd @xmath165 for the heavy target is more suppressed at low @xmath181 by multiple scatterings .
now let us take a ratio of the cross section of j/@xmath1 in pa collisions to that in pp collisions , which is called nuclear modification factor @xmath182 .
we expect that model uncertainties cancel out to some extent in the ratio .
we define @xmath182 for j/@xmath1 in our model as @xmath183 here we set the number of nucleon - nucleon collisions in pa to @xmath184 because the ugd @xmath185 scales as @xmath186 at large @xmath67 . in fig .
[ fig : rpa_jpsi_rhic ] we compare the model results for @xmath182 at @xmath187 gev with the data of @xmath188 .
note that the projectile is different between the model calculation and the data .
the notations are the same as in fig .
[ fig : jpsi_pt_rhic_pp ] .
we stress here that @xmath189 is indeed little dependent on the choice of the quark mass and factorization scale .
unfortunately , however , one immediately recognizes an unphysically strong cronin peak in the model calculations with set g1118 both at mid- and forward rapidities , which is obviously caused by the dip seen in the pp collisions ( fig .
[ fig : jpsi_pt_rhic_pp ] ) .
in contrast , the @xmath189 result with set mv looks more reasonable ; we see a moderate cronin peak at mid - rapidity due to the multiple scatterings , while it almost disappears at forward rapidity @xmath190 by the @xmath145 evolution . in low-@xmath141 region
, we also notice a too strong suppression than the experimental data .
this would imply the importance of the fragmentation process in the formation of j/@xmath1 , which is missing in a simple cem treatment . to summarize the results at rhic energy ,
the j/@xmath1 production spectrum is sensitive to the moderate value of @xmath7 , where the initial condition for the @xmath0-evolution is set .
we have a difficulty to describe the pp data and therefore the ratio @xmath182 with the ugd g1118 constrained at @xmath14 .
in contrast the set mv gives more reasonable behavior for @xmath182 . in pa collisions the @xmath141 spectrum is better described with set g1118 at mid- and forward - rapidities .
in forward rapidity , the observed @xmath141 slope is still steeper than the model , hinting other effects such as a possible energy loss of the large-@xmath15 gluon from the proton .
actually @xmath191 of j/@xmath1 at rhic energy has been studied in several approaches ( e.g. ) with introducing nuclear parton distribution and nuclear absorption effects to a j/@xmath1 production model for pp@xcite , or with taking account of the multiple scatterings and energy loss of the projectile gluons@xcite .
now we compute the j/@xmath1 production at the lhc energy , where we expect that a wider @xmath145-evolution of ugd on the nucleus side will manifest in the quarkonium spectrum .
in fact , both @xmath7 are small ( @xmath192 ) already in mid - rapidity production of the charm pair as seen in fig .
[ fig : x2-coverage ] , and as moving to larger rapidities we can probe smaller values of @xmath145 on the nucleus side down to @xmath193 .
we show in fig .
[ fig : jpsi - pp - lhc ] the j@xmath194 cross section in pp collision at @xmath195 tev , obtained in cem from charm quark spectrum ( [ eq : cross - section - ln ] ) .
notations are the same as in the case of the rhic energy . in order to assess the uncertainty
, we again vary the charm quark mass from @xmath168 to 1.5 gev , and change the factorization scale from @xmath169 to @xmath170 in the collinear approximation .
the observed data@xcite is fairly well reproduced with set g1118 in this @xmath141 region both at @xmath196 and @xmath197 , indicating that @xmath136-dependence is appropriately captured by @xmath0 evolution of ugd .
the @xmath141 slope in the collinear approximation ( [ eq : cross - section - ln - coll ] ) with set g1118 seems to be slightly off the data , while the full result with set mv gives harder @xmath141 spectrum .
the situation is expected to be similar in pp collisions at @xmath198 tev .
results in pa collisions at @xmath198 tev are plotted at mid- and forward - rapidities in fig .
[ fig : jpsi_pt_lhc ] .
the mv initial condition gives a harder spectrum of j/@xmath1 than g1118 .
but their @xmath141 slopes become almost the same at @xmath199 gev , hinting the same bfkl tail of ugd generated during the evolution .
compared to the case at the lower energy @xmath167 gev , the collinear approximation ( with set g1118 ) results in the spectral shape rather similar to the full result at this energy @xmath6 tev , where the collinear approximation on the proton side would be more suitable since the saturation scale of the nucleus is much larger than that of the proton : @xmath200 , especially in the forward region .
we show in fig .
[ fig : rpa_jpsi_lhc ] the ratio @xmath182 of j/@xmath1 as a function of @xmath201 at @xmath6 tev .
we have assumed @xmath202 as mentioned before .
we find that each band almost collapses into a single line , which means that the ratio @xmath182 is insensitive to the variation of the charm quark mass ( and the factorization scale in the collinear approximation ) within the range considered here . at mid - rapidities ( fig .
[ fig : rpa_jpsi_lhc ] ( a ) ) , we see that the ratio @xmath182 of j/@xmath1 production is suppressed at low @xmath141 , while it approaches unity at higher @xmath141 for both sets of g1118 and mv . in the collinear approximation on the proton side
, @xmath182 shows a cronin - like peak around @xmath203 gev and remains larger than unity at larger @xmath141 , which largely reflects `` @xmath182 for @xmath165 '' at the gluon level . at forward
rapidities ( fig .
[ fig : rpa_jpsi_lhc ] ( b ) ) , however , this difference due to different ugd sets and approximations becomes much weaker to yield a systematic suppression as a function of @xmath141 for all three cases .
we examine the initial - scale ( @xmath204 ) dependence of the ratio @xmath182 in fig .
[ fig : rpa_jpsi_lhc_46 ] , by plotting the results with the saturation scale @xmath205 ( upper ) and @xmath206 ( lower ) in eq .
( [ eq : initialcondition ] ) at @xmath103 .
it is found that the @xmath204 dependence of @xmath182 is relatively weak within this range . at low @xmath141
we have strong suppression , but one should keep in mind that this suppression may be filled to some extent by the nonperturbative fragmentation in forming j/@xmath1 , as is inferred from the discussion on fig . [
fig : rpa_jpsi_rhic ] . to summarize the result at lhc energy
, we can probe here a wide @xmath145-evolution of the ugd @xmath207 through the j/@xmath1 production , and the ratio @xmath182 will be a good indicator for it .
next we consider @xmath2 production .
non - linear effects are generally suppressed by the inverse power of the heavy quark mass . however , since the bottom quark mass @xmath208 is just three times as heavy as the charm quark mass @xmath10 , the relevant value of @xmath0 for the @xmath2 production becomes larger by the same factor at low @xmath141 , as compared to the j/@xmath1 . at the lhc energy , this @xmath0 value may be still small enough for multiple scatterings and saturation to be relevant in the @xmath18 production .
we plot the @xmath141 spectrum of @xmath2 in pp and pa collisions at @xmath209 and 5.02 tev , respectively , in figs .
[ fig : upsilon - pp ] and [ fig : upsilon - pt - lhc ] , together with the data measured by atlas and lhcb@xcite for the pp case .
here we have chosen the cem parameter as @xmath210 , and varied @xmath208 from 4.5 to 4.8 gev .
other notations are the same as in the j/@xmath1 case . in pp collisions ,
the coincidence between the model and the data for @xmath2 state is not as good as that for j/@xmath1 at low @xmath141 and at forward rapidity .
we present in fig .
[ fig : rpa_upsilon_lhc ] the nuclear modification factor @xmath182 for @xmath2 as a function of @xmath201 .
the model uncertainty from the quark mass value and the factorization scale would cancel out by taking the ratio of the cross - sections in the pp and pa collisions .
indeed , each band collapses into a thin line whose width is almost unnoticeable .
this result for @xmath2 is qualitatively very similar to that for j/@xmath1 . at mid - rapidity
, we see a suppression @xmath182 in low @xmath141 region below 5 gev , while it turns back to unity at larger @xmath141 . only in the collinear approximation
, we see the cronin - like enhancement , which is largely caused by the dip structure in the proton ugd at moderate @xmath15 . at forward
rapidities @xmath211 , the @xmath18 production is suppressed in a wide @xmath141 region from 0 to 20 gev , irrespective of the model ugd s , g1118 or mv , or of the use of collinear approximation . in the forward region ,
@xmath2 production has the sensitivity to the small-@xmath0 evolution of ugd in the nucleus .
we have also checked the initial - scale ( @xmath204 ) dependence of @xmath182 for @xmath2 by comparing the result with @xmath212 and @xmath213 to find that the change is very similar to the case with j/@xmath1 ( fig .
[ fig : rpa_jpsi_lhc_46 ] ) .
we study the rapidity dependence of the ratio @xmath182 integrated over @xmath141 .
the computation is performed with set g1118 . in fig .
[ fig : rpa_jpsi_y_dep ] shown is @xmath214 of j/@xmath1 at @xmath215 and 5.02 tev , together with that of @xmath2 for the latter .
note that our assumption of dilute - dense colliding system applies only in the positive rapidity region ( @xmath216 ) , especially for pp , which is needed in the denominator of @xmath182 .
we see systematically a stronger suppression of @xmath182 as the rapidity increases both at rhic and lhc energies .
this is in accord with @xmath0-evolution of ugd in the heavy target .
@xmath214 of j/@xmath1 flattens out at @xmath217 at rhic energy because the j/@xmath1 is produced there by the gluons with @xmath218 and we freeze the saturation scale to its initial value @xmath219 . comparing the results of j/@xmath1 and @xmath2 at lhc
, we note that the suppression of @xmath2 is smaller than that of j/@xmath1 , but is still significant to be observed
. it would be quite important to study these systematics in experimental data in order to quantify the saturation effects in the heavy nuclear target .
it would be interesting to study the dependence of @xmath182 on the saturation scale parameter @xmath204 , which may be translated to the effective thickness of the target .
we compute @xmath182 of j/@xmath1 and @xmath2 integrated over @xmath141 as a function of @xmath204 at several values of @xmath136 .
we fix here the ugd set g1118 and the quark masses as @xmath220 gev and @xmath221 gev . in fig .
[ fig : rpa_qs_jpsi ] we plot @xmath182 of j/@xmath1 at @xmath215 and 5.02 tev .
we found that for each rapidity @xmath204-dependence of @xmath182 can be fitted nicely by a model function : @xmath222 with @xmath114 , @xmath223 and @xmath224 being parameters .
this functional form is motivated by qcd analog of _ superpenetration _ of a electron - positron pair through a medium@xcite .
we show @xmath2 result in fig . [
fig : qs - upsilon ] with fitted curves . the stronger suppression at the larger value of @xmath204 is naturally understood as a result of stronger multiple scatterings and saturation effects in the heavier target .
energy and rapidity dependences may be qualitatively inferred through the increase of @xmath225 with increasing @xmath136 .
thus we tried to fit the rapidity dependence of @xmath182 by replacing in eq .
( [ eq : rpa - fit ] ) @xmath226 with a free parameter @xmath227 , but it was only unsatisfactory .
we remark here that quarkonium suppression due to parton saturation in our treatment is twofold : a relative depletion of the gluon source and multiple scatterings of the quark pair in the target .
the latter disturbs the boundstate formation , by increasing the pair s invariant mass on average in cem@xcite .
it appears hard to describe energy and rapidity dependence of the suppression at the same time through a single function @xmath228 .
finally , we study the mean transverse momentum of quarkonium in pa collisions .
the momentum broadening in the nuclear target has been discussed in the literature@xcite . in our framework , the multiple scatterings of the incident gluon and the produced quark pair in the nuclear target , encoded in @xmath33 and @xmath229 terms in eq .
( [ eq : mf - final-1 ] ) respectively , cause the momentum broadening of the pair .
typical momentum transfer of the multiple scatterings in the nucleus should be characterized by the saturation scale @xmath230 .
we define here the broadening of @xmath141 as the deviation of the mean transverse momentum squared @xmath231 of j/@xmath1 in pa collisions from that in pp collisions : @xmath232 in fig .
[ fig : pt_qs_jpsi ] we plot @xmath233 as a function of @xmath204 .
we use ugd set g1118 with the quark masses @xmath220 gev and @xmath221 gev .
we have found that for each rapidity the @xmath204 dependence of the broadening can be fitted in a simple form : @xmath234\end{aligned}\ ] ] with @xmath114 and @xmath224 being parameters . at @xmath167 gev ,
the broadening at mid - rapidity is obviously linear in @xmath204 , which indicates the random walk nature of the multiple scatterings in the momentum space . in the forward region
, we naively expected an increase of the mean momentum by the stronger scatterings , but actually found the opposite , i.e. , a decrease from the mid - rapidity value .
we interpret this as the effect of kinematical boundary of @xmath15 in the forward region ( see fig . [
fig : x2-coverage ] ) .
the measured value of @xmath235 at rhic@xcite seems to be smaller by a factor of 5 than that in fig .
[ fig : pt_qs_jpsi ] , if we naively translate @xmath204 to the centrality parameter @xmath236 evaluated for dau collisions .
this strong broadening originates probably from the fact that our model has too hard @xmath141 spectrum at rhic energy .
but it is at least consistent with data that @xmath141 broadening at forward rapidities @xmath237 is weaker than that at mid - rapidity @xmath238 . at @xmath6 tev ,
a wider phase space opens up and we instead see an increase of the mean momentum of j/@xmath1 as moving to the forward rapidity region .
we have checked that @xmath233 gets back to be smaller at @xmath239 than that of mid - rapidity , just as seen in the case of @xmath167 gev .
nonlinear dependence on @xmath204 may imply the different evolution speed of multiple - scattering strength for different initial values @xmath204 .
the result for @xmath2 in fig .
[ fig : upsilon - pt - qs ] is similar to the j/@xmath1 case , but interestingly the broadening becomes more remarkable ; the heavier bottom quark pair can acquire the larger transverse momentum @xmath141 in multiple scatterings before going beyond the threshold set on the pair s invariant mass @xmath240
quarkonium production in proton - lead collisions provides us with a good opportunity to study the saturation phenomenon in the incident nucleus thanks to the wide kinetic reach at the lhc .
we have computed the j/@xmath1 and @xmath2 production in pa collision at collider energies within cem based on the cgc quark pair production , and have discussed sensitivity of the quarkonium observables to the parton saturation in the target nucleus .
we have presented the calculations with the ugd set g1118 which is constrained with dis data at @xmath109 , with and without the collinear approximation , and the calculations with the model ugd set mv for comparison . at the rhic energy @xmath167 gev ,
the j/@xmath1 at mid - rapidity is produced not from the small-@xmath0 gluons , but from the moderate-@xmath0 gluons , and the @xmath141 spectrum in pp collisions is unfortunately sensitive to a unphysical dip structure of the ugd set g1118 , which was constrained only for @xmath104 .
we need better extrapolation of our framework to @xmath241 . in pa collisions ,
multiple scatterings smear out the dip of the ugd and the @xmath141 spectrum of j/@xmath1 becomes closer to the observed one in dau collisions . at the lhc energy @xmath6 tev
, the small-@xmath0 gluons dominate the charm production , and we have found that our model with the ugd set g1118 works for j/@xmath1 production in pp collisions both at mid- and forward - rapidities .
then we have shown our model prediction on j/@xmath1 production in pa collisions .
the ratio @xmath242 for j/@xmath1 shows a suppression for @xmath243 gev at mid - rapidity due to saturation effects , and it is further suppressed in wider range of @xmath141 as moving to forward rapidities .
we have also shown that the @xmath2 production in pa collisions at the lhc has a good sensitivity to the gluon saturation of the nucleus , provided that the effect is smaller than that in the j/@xmath1 case . in our model , when integrated over @xmath141 , the ratio @xmath214 for @xmath18 at the lhc shows a suppression similar to that of j/@xmath1 at rhic energy .
transverse momentum broadening @xmath244 of the quarkonium shows an increasing behavior as a function of @xmath204 .
because our model gives harder @xmath141 spectrum than the data , the broadening is likely to be overestimated at rhic energy in our calculation .
however , it is still interesting to notice that at rhic energy the broadening becomes weaker at forward rapidity than at mid - rapidity due to the kinematical constraint . at lhc energy , on the other hand , we expect the increase of the broadening at forward rapidities because of the larger saturation scale @xmath245 and wider kinematic coverage of the lhc .
transverse momentum broadening is also investigated recently by taking account of the multiple scatterings in the target in @xcite . in conclusion
, we have numerically studied the quarkonium production in pa collisions at the rhic and lhc , within cem based on the cgc quark - pair production energies , and have quantifies the effects of saturation and multiple scatterings in the target nucleus on the j/@xmath1 and @xmath18 observables .
comparison of our results with experimental data at the lhc must be very important to access the relevance of the saturation physics in the quarkonium production .
in this work we have employed cem to describe the nonperturbative formation of the quarkonia .
in fact , quarkonium formation is one of the challenges in qcd , even in pp collisions , and cem replaces this just by a probability constant @xmath246 .
non - relativistic qcd framework has recently been extended to nlo , which improves our understanding of quarkonium production significantly@xcite . as a first step in this direction we plan to match the quark - pair production from cgc onto the non - relativistic qcd approach .
we can investigate also the production of open heavy flavor mesons in our framework .
modification of @xmath141 spectrum and @xmath247 correlations will also provide very useful information on the saturation in the target nucleus and a good benchmark for the energy loss and collective flow measurements of @xmath154 and @xmath248 mesons .
we will report this elsewhere@xcite .
we are very grateful to j. albacete , a. dumitru , f. gelis , k. itakura , y. nara , r. venugopalan for useful discussions and collaborations on related topics .
this work was partially supported by grant - in - aids for scientific research ( ( c ) 24540255 ) of mext . | quarkonium production in high - energy proton ( deuteron)-nucleus collisions is investigated in the color glass condensate framework .
we employ the color evaporation model assuming that the quark pair produced from dense small-@xmath0 gluons in the nuclear target bounds into a quarkonium outside the target . the unintegrated gluon distribution at small bjorken @xmath0 in the nuclear target is treated with the balitsky - kovchegov equation with running coupling corrections .
for the gluons in the proton , we examine two possible descriptions , unintegrated gluon distribution and ordinary collinear gluon distribution .
we present the transverse momentum spectrum and nuclear modification factor for j/@xmath1 production at rhic and lhc energies , and those for @xmath2 at lhc energy , and discuss the nuclear modification factor and the momentum broadening by changing the rapidity and the initial saturation scale .
institute of physics , university of tokyo , + komaba 3 - 8 - 1 , tokyo 153 - 8902 , japan |
quantum discord has proved to be a resource for performing some quantum information and computation protocols @xcite .
this approach of quantum correlation expresses a different type of correlation @xcite in comparison with entanglement , in which we are concerned about separability .
quantum discord sheds a new light on the concept of correlation in composite systems and reveals that there are some separable , i.e. disentangled , multipartite states which possess quantum correlation and hence they can be employed as a resource for improving the quantum information processes , speeding up the quantum computation algorithms and/or performing the quantum communication protocols . despite the fact that entanglement looks at quantum correlation from the separability point of view
, quantum discord can capture quantum correlation from the measurement perspective @xcite .
however , calculation of quantum discord requires finding the best ( optimised ) measurement protocol to be performed on one part of the system , therefore , complicating the calculation of quantum discord and prevents one from obtaining an analytic closed - form of the formula for quantum discord in general .
indeed , quantum discord is analytically calculated only for a few families of two - qubit states @xcite , for some reduced two - qubit states of pure three - qubit states , and also for a class of rank-2 mixed state of @xmath0 systems @xcite . on the other hand ,
the geometric measure of quantum discord is another measure of quantum correlation , that is in general easier to be utilised for calculations of non - classical correlations @xcite .
furthermore , dakic _
et al_. have demonstrated , by using a variety of polarisation - correlated photon pairs , that non - zero quantum discord is the necessary resource for remote state preparation @xcite .
also , they show that the geometric measure of quantum discord is directly linked to the fidelity of remote state preparation for a broad class of states . this fact may provide an operational interpretation for this measure of quantumness .
geometric discord is defined as the squared hilbert - schmidt distance between the state of the quantum system and the closest zero - discord state . for a bipartite state @xmath1 on the hilbert space @xmath2 , geometric discord @xmath3
is defined as @xcite @xmath4 where minimisation is taken over the set of all zero - discord states @xmath5 , and @xmath6 is the squared norm in the hilbert - schmidt space .
an exact expression of @xmath7 for pure @xmath8 and arbitrary @xmath9 states are obtained @xcite .
consider a bipartite state @xmath1 acting on the hilbert space @xmath10 .
one can write @xmath1 in the bloch representation as @xmath11 where @xmath12 are the usual pauli matrices , and @xmath13 are generators of @xmath14 , fulfilling the following equations @xmath15 also @xmath16 with @xmath17 \label{29}$ ] , and @xmath18 with @xmath19 $ ] denote local coherence vectors of two subsystems , and @xmath20 with @xmath21 \label{31}$ ] is the correlation matrix of the state .
it is shown that @xmath7 of @xmath22 states can be written as @xcite @xmath23 where @xmath24 are eigenvalues of @xmath25 , in non - increasing order , and @xmath26 . the jaynes - cummings model ( jcm ) , which describes the interaction of a single two - level atom with the quantised radiation field @xcite , has been the subject of numerous studies in the field of quantum optics due to the fact that this system shows behaviour which clearly illustrates physical concepts such as collapse and revival .
furthermore , it has attracted much attention for the study of quantum correlations .
a lot of studies have been devoted to understand the atom - field interaction and methods have been devised to create , control , and quantify quantum correlation between them both in the weak and strong coupling regimes @xcite . in this paper , we specifically intend to investigate the quantum correlation of a system which exhibits decoherence .
decoherence is an inevitable consequence when the quantum system is interacting with a surrounding environment .
if energy transfer occurs between the system and environment , the dissipation process is the most important mechanism leading to decoherence .
however , decoherence can happens even when energy is conserved in the system : in this case the process of phase damping ( dephasing ) reduces the quantum coherence of the system .
considering the properties of environment is important to investigate its interaction with the quantum systems and decoherence ; the concepts of markovian and non - markovian processes that deal with the capability of the environment to possess memory of the system have been the subject of many studies . for example , the dynamical behavior of quantum correlation of bipartite systems is reviewed in @xcite where the authors compared the behavior of entanglement and quantum discord in non - markovian and markovian processes .
moreover , milburn has shown that , even without explicitly model an environment and its interaction with the system , the dephasing process may occur due to quantum jumps in the evolution of the system @xcite .
investigating the quantum discord of open systems has been the subject of several studies .
for example , jara @xmath27 have studied the discording power of an interaction between a system and its enclosing bath , both with and without the assumption of the rotating - wave approximation ( rwa ) @xcite .
in addition , altintas @xmath27 have used the master equation of evolution of interacting atom and field and investigated the entanglement and quantum discord of atom - field interactions in strong coupling regime .
also , dynamics of the geometric quantum discord was studied in @xcite , where the authors have employed a dissipative system of two independent atom - cavity - reservoirs , both in the strong and weak coupling regimes . in the present paper , we investigate the dynamical behavior of quantum correlations of a system consisting of a two - level atom which interacts with a single - mode quantised radiation field .
such systems are important in operational quantum information processing .
in particular , this system is suitable for performing some quantum communication tasks , because it includes a stationary qubit ( atom ) and a flying qudit ( photon ) .
the atom - field interaction is described by the jaynes - cummings model .
we use the jcm in the weak coupling regime together with the rwa , and assume that our system undergoes a phase damping evolution caused by a decoherence process .
the physics of this process could be quantum jumps of the atom - field state during the evolution .
we focus on calculating the time - evolved state of the atom - field , where the atom is initially in a general mixed state and the field is initially in a pure state , i.e. number and coherent states , and investigate the quantum correlations between the atom and field . our algebraic method to calculate the geometric discord of the jcm
is novel , in which the calculation of coherence vectors and correlation matrix in the jcm are carried out easily . in previous studies such as @xcite the pure dephasing evolution of the jcm has been considered , i.e. when the atom and field are both initially in pure states , but we consider here a more general case , when the initial state of the atom is a general mixed state .
also , our analytic results enable us to evaluate the dynamics of the quantum correlations of the jcm such as negativity and geometric discord for various values of atomic mixedness of the system , and the atom - field detuning .
our results reveal that although the asymptotic geometric discord of the system is negligible in the resonance and far resonance regions , it has a non - zero value for some intermediate detunings , and that we can find an optimum value for such asymptotic @xmath28 in an intermediate value of detuning .
interestingly , a similar result is mentioned in @xcite for the negativity of the jcm and we show here how these two types of quantum correlation resemble each other in this view .
the paper is organised as follows . in section
ii , we briefly review the jaynes - cummings model .
pure dephasing evolution of the system is studied in section iii .
section iv includes the calculation of the geometric discord for the two distinct initial states of the field i.e. number and coherent states . in section
v , we present our results and conclude the paper with some discussions .
the jaynes - cummings model ( jcm ) describes the interaction of a two - level atom with a single - mode quantised radiation field @xcite .
the hamiltonian of this model is given by @xmath29 where @xmath30 is the internal hamiltonian of the system and @xmath31 describes the atom - field interaction in the rotating - wave approximation ( rwa ) and weak - coupling regime , where ( for @xmath32 ) @xmath33 here @xmath34 and @xmath35 are the transition frequency of the two - level atom and the frequency of the radiation field , respectively , and @xmath36 denotes the atom - field coupling .
note that the rwa and weak coupling regime are valid when @xmath37 .
also @xmath38 , @xmath39 are atomic spin - flip operators , and @xmath40 is the atomic inversion operator which act on the atom hilbert space @xmath41 .
also @xmath42 and @xmath43 are annihilation and creation operators of the field acting on the field hilbert space @xmath44 .
the two - dimensional hilbert space of the atom is spanned by two orthonormal states @xmath45 and @xmath46 , and the field hilbert space is spanned by the photon number states @xmath47 , where @xmath48 is the vacuum state of the radiation field . to achieve the time evolution of the system , it is convenient to use the dressed - state representation of the hamiltonian .
regarding the fact that hamiltonian conserves the total number of excitations of the operator @xmath49 , one can decompose the atom - field hilbert space @xmath50 as @xmath51 such that @xmath52 and @xmath53 for @xmath54 , are the eigen - subspaces of @xmath55 corresponding to the eigenvalues @xmath56 and @xmath57 , respectively .
accordingly , hamiltonian has the eigenvalues @xmath58 with the corresponding eigenvectors @xmath59 where @xmath60 , and @xmath61 is the rabi frequency .
also @xmath62 is the detuning parameter of the system .
decoherence destroys the quantumness of the system and decreases the useful quantum correlations between on the parts of the system .
there are several approaches to consider decoherence in a quantum system which are responsible for quantum - classical transition .
one of these approaches is based on the modified schrdinger equation in such a way that the quantum coherence is automatically destroyed as the system evolves .
this mechanism is called _ intrinsic decoherence _ and has been studied in the framework of several models ( see @xcite and references therein ) . in particular , milburn has proposed a simple modification of the standard quantum mechanics based on the assumption that for sufficiently short time steps @xmath63 the system evolution is governed by a stochastic sequence of identical unitary transformations rather than a continuous unitary evolution @xcite .
this assumption leads to a modification of schrdinger equation which includes a term corresponding to the decay of quantum coherence in the energy bases . using a poisson model for the stochastic time steps , milburn obtained the following dynamical master equation in the first order approximation @xcite @xmath64-\frac{\gamma}{2}[h,[h,{{\rho ( t)}}]],\ ] ] where @xmath65 is the density matrix of the atom - field in any given time @xmath66 .
equation has the following formal solution @xcite @xmath67 such that @xmath68 is the initial state of the system and @xmath69 in order to find the time - evolved state @xmath65 , we need to expand the initial state @xmath68 in terms of the dressed - states , i.e. the hamiltonian eigenbases , as @xmath70 inserting this in eq . we get @xmath71 where we have defined @xmath72 now let us suppose that , initially at @xmath73 , the system is found in the product state @xmath74 such that @xmath75 is the initial state of the atom and is considered to be a mixed state @xmath76 and @xmath77 , the initial state of the field , is assumed to be a pure state @xmath78 with @xmath79 , where the complex coefficients @xmath80 satisfy the normalisation condition @xmath81 . in section [ gd section ] , we will fix the coefficients @xmath82 for two special cases , namely number states and coherent states . accordingly , inserting eqs . - into eq . and using orthonormal bases @xmath83 for the atomic hilbert space , one can find , after tedious but straightforward calculations , the following representation for the density matrix @xmath84 @xmath85 where @xmath86 , @xmath87 , and @xmath88 , operators acting on the field hilbert space @xmath44 , are defined by @xmath89 where matrix elements of operators @xmath90 , @xmath91 , and @xmath92 in the fock bases @xmath93 are given by @xmath94\nonumber \\ & + \cos^{2}(\theta_{n})\big[\sin^{2} ( \theta_{m})\exp\big(-i\omega^{nm}_{-+}t-\frac{\gamma t}{2}(\omega^{nm}_{-+})^{2}\big)\nonumber \\ & + \cos^{2} ( \theta_{m})\exp\big(-i\omega^{nm}_{--}t-\frac{\gamma t}{2}(\omega^{nm}_{--})^{2}\big)\big]\bigg],\end{aligned}\ ] ] @xmath95,\end{aligned}\ ] ] @xmath96,\end{aligned}\ ] ] @xmath97\nonumber \\ & + \sin^{2}(\theta _ { n-1})\big[\cos^{2}(\theta_{m-1})\exp\big(-i\omega^{n-1,m-1}_{-+}t-\frac{\gamma t}{2}(\omega^{n-1,m-1}_{-+})^{2}\big)\nonumber \\ & + \sin^{2}(\theta_{m-1})\exp\big(-i\omega^{n-1,m-1}_{--}t-\frac{\gamma t}{2}(\omega^{n-1,m-1}_{--})^{2}\big)\big ] \bigg],\end{aligned}\ ] ] @xmath98\nonumber \\ & + \cos^{2}(\theta_{n})\big[\exp\big(-i\omega^{n , m-1}_{-+}t-\frac{\gamma t}{2}(\omega^{n , m-1}_{-+})^{2}\big)\nonumber \\ & -\exp\big(-i\omega^{n , m-1}_{--}t-\frac{\gamma t}{2}(\omega^{n , m-1}_{--})^{2}\big)\big]\bigg],\end{aligned}\ ] ] @xmath99\nonumber \\ & + \sin^{2}(\theta_{m-1})\big[\exp\big(-i\omega^{n , m-1}_{+-}t-\frac{\gamma t}{2}(\omega^{n , m-1}_{+-})^{2}\big)\nonumber \\ & + \exp\big(-i\omega^{n , m-1}_{--}t-\frac{\gamma t}{2}(\omega^{n , m-1}_{--})^{2}\big)\big]\bigg].\end{aligned}\ ] ]
in this section , we turn our attention on the geometric quantum discord of the state given by eq . .
to do so , we should first note that the state is supported , actually , on a @xmath100 hilbert space and investigating the correlation properties of this system may appear to an impossible task .
however , as we will show in the following sections , by considering some appropriate initial states for the field , our state at a given time @xmath101 can be represented by a @xmath102 matrix for a finite @xmath103 . in continue , we express two examples of these preparations .
we first consider the case that the field is prepared , initially , in a given number state @xmath104 , i.e. the coefficients @xmath105 are set to be @xmath106 . in this particular case , the atom - field state @xmath84 of eq .
is supported on a @xmath107 hilbert space . by setting @xmath108 as the orthonormal bases of the field , one can represent the corresponding density matrix as @xmath109 where @xmath110 this density matrix can be used to obtain the coherence vector @xmath111 and correlation matrix @xmath112 , and hence the geometric discord . as the second example
, we consider the case that the field is initially in a coherent state @xmath113 , i.e. @xmath114 . although this initial condition implies a @xmath100 support for @xmath84 , the poissonian distribution of photon numbers in coherent states allows us to truncate the dimension of the field hilbert space to a finite one .
numerically , it is seen that if we set @xmath115 the coefficients @xmath105s will be negligible for @xmath116 ; indeed , the ratio @xmath117 becomes less than @xmath118 for large enough values of @xmath119 , so it is reasonable to restrict the dimension of the field hilbert space to an appropriate @xmath103 . to fulfill the adequate accuracy in the case of @xmath115 ,
numerical calculations have shown that @xmath120 is sufficient , such that @xmath121\cong 1 $ ] , so we truncate @xmath44 to a 30-dimensional space .
this approach was also employed in @xcite to calculate the negativity of the jcm where the system undergoes a unitary evolution , and the consideration of the initial states of the atom and field are the same as the present paper .
we use the same approach to evaluate the dephasing evolution of the jcm , and it can be seen that just by setting the dephasing parameter @xmath63 equal to zero , i.e. the unitary evolution , the results of @xcite for the state of the system can be achieved .
nevertheless , in the case of dephasing evolution , the 25-dimensional field hilbert space which is used in @xcite is no longer accurate in numerical calculations , so we have expanded this dimension to thirty .
now we focus on calculating the @xmath7 of the atom - field state while the field is initially in the mentioned coherent state .
consequently , each block of the density matrix will be a @xmath122 matrix acting on the field .
if we let @xmath123 to be the generators of @xmath124 and use the pauli matrices as the generators of the @xmath125 , and utilise the representation in , we have @xmath126 , \\ x_{2}(t)&={{\mathrm { tr}}}_{f } \big[-i\hat{c}^{\dagger}(t)+i\hat{c}(t)\big ] , \\ x_{3}(t)&={{\mathrm { tr}}}_{f } \big[\hat{a}(t)-\hat{b}(t)\big],\end{aligned}\ ] ] and @xmath127 , \\
t_{2j}(t)&=15{{\mathrm { tr}}}_{f } \big[\hat{\lambda}_{j}\big(-i\hat{c}^{\dagger}(t)+i\hat{c}(t)\big)\big ] , \\ t_{3j}(t)&=15{{\mathrm { tr}}}_{f } \big[\hat{\lambda}_{j}\big(\hat{a}(t)-\hat{b}(t)\big)\big],\end{aligned}\ ] ] where @xmath128 and @xmath129 is the partial trace over the field which can be readily applied on the matrix representation of the operators @xmath86 , @xmath87 , @xmath88 and @xmath130 . having the coherence vector @xmath111 and correlation matrix @xmath112 in hand
, one can calculate the geometric discord via eq . .
meanwhile , an explicit representation of the generators of @xmath124 is required ; the method proposed in @xcite has expressed a set of straightforward equations to achieve both cartan sub - algebra and non - diagonal generators of @xmath14 .
based on the analytic calculations of the pure dephasing evolution of the atom - field system expressed in the previous sections , we have calculated the geometric discord @xmath7 of the state for different values of the parameters @xmath36 , @xmath131 , @xmath63 and @xmath132 .
the evolution of the state of the system and hence the dynamics of the quantum correlations strongly depend on the initial state of the system .
therefore , in all of the following calculations we suppose that the atom can be found , initially , in a mixture of the ground and excited states , and the field is in a pure initial state . for the pure state of the field
we consider two distinct cases : the number state and the coherent state .
we mostly focus our attention on the quantum correlations of the jcm in pure dephasing evolution , but it is worthwhile to evaluate our results for @xmath133 , i.e. the unitary evolution , and then compare them with the results of the situation when the dephasing process is present .
we first consider the case that the field is initially prepared in a number state , so that the behaviour of the system over time can be readily extracted by examining eq . .
* _ unitary evolution ( @xmath133 ) _ + suppose that the initial state of the field is set to be vacuum state @xmath134 .
figure [ fig1]-(a ) shows the time behaviour of @xmath7 when the atom is initially in the excited state @xmath135 , and for some regular values of detuning @xmath131 , where rwa is valid .
+ and negativity when the field is initially in the vacuum state @xmath134 , and @xmath133 ( unitary evolution ) , @xmath136 , @xmath137 with @xmath138 [ solid black line ] and @xmath139 [ dashed red line].,title="fig:",width=680 ] + [ fig1 ] + since evolution is unitary , the pure initial state of the system will remain pure during the evolution . as it is mentioned in ref .
@xcite , for a @xmath140 pure state @xmath141 , the geometric discord is related to the generalised concurrence as @xmath142 where @xmath143 is the generalised concurrence of @xmath144 @xcite .
figure [ fig1]-(b ) shows the dynamics of the negativity of the jcm with the same assumptions .
as it is clear , the time behaviour of @xmath7 and negativity are the same up to a scale factor .
we can see that as the collapse and revival occurs , the atom and the field becomes periodically correlated and de - correlated . *
_ dephasing regime ( @xmath145 ) _ + when the dephasing parameter @xmath63 is not zero , the story is different .
decoherence suppresses the coherence oscillations ( collapse and revival ) of the evolution and after sufficient time , it leads the system to get a stationary state . in figs .
[ fig2 ] and [ fig3 ] we have plotted @xmath3 versus time for different values of @xmath132 and @xmath131 when the initial state of the field is @xmath146 .
+ and negativity when field is initially in @xmath146 for @xmath147 , @xmath148 and @xmath149 with @xmath138 [ solid black line ] , @xmath150 [ dashed red line ] and @xmath151 [ dot - dashed blue line].,title="fig:",width=680 ] + and negativity when field is initially in @xmath146 for @xmath137 , @xmath148 and @xmath136 with @xmath138 [ solid black line ] , @xmath150 [ dashed red line ] and @xmath151 [ dot - dashed blue line].,title="fig:",width=680 ] + since elements of the density matrix are proportional to @xmath152 , they vanish after times @xmath153 , for @xmath154 .
so the system reaches a stationary state , asymptotically .
hence we have the following time - independent elements for the density matrix of the system at the asymptotically large times : @xmath155 as it is clear , the detuning parameter @xmath131 have an important role in the asymptotic geometric discord @xmath156 .
for instance , for @xmath138 we have @xmath157 , @xmath158 and @xmath159 i.e the density matrix becomes diagonal and consequently @xmath160 .
so it seems that we need a nonzero @xmath131 in order to have nonzero @xmath161 .
likewise , the evolution of the system barely creates any correlation for large values of detuning .
accordingly , the asymptotic geometric discord @xmath161 may have a maximum value at a finite optimum value of detuning , @xmath162 . figure [ fig4 ] expresses the asymptotic geometric discord @xmath161 versus @xmath131 when the field is initially in @xmath163 and for different values of @xmath132 . finally , figure [ fig5 ] summarise the previous results in a 3d plot of @xmath161 versus @xmath131 and @xmath132 .
+ as a function of @xmath131 when the field is initially in @xmath146 and @xmath137 with @xmath164 [ solid black line ] , @xmath149 [ dot - dashed red line ] and @xmath136 [ dashed blue line].,title="fig:",width=377 ] + these figures reveal that the value of @xmath165 seems to be not symmetric with respect to the parameter @xmath132 .
since the one - dimensional dressed state @xmath166 does not change through the evolution , it turns out that the rules of @xmath132 and @xmath167 differ in evolution , leading therefore to the above mentioned asymmetry .
+ as a function of @xmath131 and @xmath132 when the field is initially in the number state @xmath146.,title="fig:",width=453 ] we now check our results in the case where the field is prepared initially in a coherent state @xmath113 . in the following case ,
we fix @xmath115 , so it is sufficient to use only the 30-dimensional truncated subspace of the field fock space . *
_ unitary evolution ( @xmath133 ) _
+ first , suppose the evolution is unitary , i.e. @xmath133 . figure [ fig6 ] shows @xmath7 of the jcm when the atom is initially in the excited state , i.e. @xmath136 , and in resonance with the field , i.e. @xmath138 . authors of @xcite have calculated the negativity and mutual information of the jcm with the similar assumptions . evidently , apart from a difference in scaling , fig .
[ fig6 ] exhibits coincidence of the results : the dynamical behavior of @xmath7 coincides with the dynamical behavior of negativity presented in @xcite .
+ when field is initially in @xmath113 for @xmath133 ( unitary evolution ) , @xmath137 , @xmath138 and @xmath136 with @xmath115.,title="fig:",width=453 ] * _ dephasing regime ( @xmath168 ) _ + the dephasing mechanism tends to demolish the coherent elements of the density matrix eq .
( suppose @xmath114 for @xmath169 ) .
it is evident that for the asymptotically large times ( @xmath153 ) , the coherent term of the density matrix , @xmath170 , approaches zero unless in the cases where @xmath171 ; this happens only for eight elements , namely four diagonal and four non - diagonal elements . for large times , however , it is not difficult to see that under the resonant condition @xmath138 , only the diagonal elements remain and consequently the density matrix becomes zero - discord . in the opposite situation ,
when the detuning @xmath131 is large enough , the non - diagonal elements vanish for the asymptotically large times .
similar to the conclusion in the previous case , there exists an optimum value of detuning which gives the maximum value for the geometric discord of the stationary state .
figure [ fig7 ] shows the asymptotic geometric discord @xmath161 versus @xmath131 when the initial state of the field is in the coherent state @xmath113 with @xmath115 .
+ as a function of @xmath131 when field is initially in the coherent state @xmath113 , with @xmath164 [ solid black line ] , @xmath149 [ dot - dashed red line ] and @xmath136 [ dashed blue line].,title="fig:",width=453 ] to discuss some aspects of our results , our calculations reveals that for the asymptotic case , i.e. when @xmath172 , the support of the density operator of the system lies only on a 4-dimensional subspace , irrespective of the initial state .
one can vary this support by changing the system parameters , for instance the detuning @xmath131 . in this case , the asymptotic geometric discord depends on @xmath131 , @xmath36 and @xmath132 ( also @xmath173 for number states ) . for instance , for the case of @xmath149 ,
the supports of the stationary states coincide for the coherent and the number initial states of the field .
this fact is obvious from figures [ fig4 ] and [ fig7 ] ; these figures are exactly the same . for other values of @xmath132 ,
the rank of the support remains unchanged but the above mentioned coincidence does not occur .
+ also , it is worthwhile to discuss how correlation can survive in an off - resonant system while they do not survive in the resonant case .
the quantum correlation can resulted either from a resonant transfer of energy between the atom and the field , or from dispersive energy shifts affecting the atoms and the photons when they are not resonant @xcite .
the fact is that , although the system gains significant amounts of correlation in the resonant case , this correlation is vulnerable to dephasing and can not survive . on the other hand ,
the quantum correlation of the system in off - resonant case is rather robust to the dephasing process and are longer - lasting . strictly speaking
, the energy transfer process dominates the resonant atom and field , while the latter dominates the far off - resonant system .
our results indicate that although the contributions of the two above processes change as we alter the detuning , the ability of the system to gain correlation is generally remarkable for detunings between zero up to some intermediate value ( notice figure [ fig1 ] ) , while this ability decreases for larger detunings ; so we can conclude that the ability of the system to gain correlation does not immediately wane by an increase in detuning , but the reduction of this ability occurs when the detuning is greater than some intermediate value .
in addition , the immunity of the system to the dephasing is very important , and it strongly depends on the initial state of the system , and also on the amount of detuning . speaking about the immunity ,
the coherent elements ( off - diagonal elements ) of expression shown in are fragile to dephasing , and vanish at large times , so their influence in maintaining correlation of the system is highly dependent on the dephasing process .
but since the energy of the atom - field system is conserved , the population elements of are invariant and their correspondence in the quantum correlation of the system can survive . we speculate that for the initial state , the resonant transfer of energy between atom and field which dominates the behaviour of the correlation in a resonant system is highly relevant to the coherent elements of , while for an off - resonant system , the amounts of correlation is more relevant to the population elements , and accordingly , is more robust against the dephasing processes .
the fact is that detuning changes the energy levels of the atom - field system and the coherent and population elements of the evolution , so it influences both the ability of system to create correlation and the immunity of correlation to the dephasing .
mathematically , consider the density matrix of to be the sum of two matrices as below @xmath174 where the @xmath175 shows the diagonal elements of @xmath84 corresponding with energy levels of hamiltonian ( 5 ) and @xmath176 corresponds with off - diagonal elements of the density matrix ( 14 ) .
expanded in detail , we have @xmath177 and since the energy of the system is conserved , this matrix is time - invariant , whereas @xmath176 vanishes at large times because of dephasing .
accordingly , we have @xmath178 .
it is evident that the off - diagonal elements of @xmath175 in the standard atom - field representation ( not dressed states ) are crucial in quantum correlation of the system , and they depend on both the initial state of the system , @xmath68 , and the parameter @xmath179 ( see ) which determines the energy levels of the system .
so for a given @xmath68 , the parameter @xmath179 governs @xmath180 and consequently determines the amounts of correlation of the system at asymptotically large times .
therefore , both the parameters @xmath131 and @xmath36 should be taken into account to study the asymptotic correlation of the system ( also @xmath119 is important when the initial state of the field is a number state ) .
with respect to our assumption for the initial state , the state @xmath181 is correlation - free when the system is in resonance , whereas , it contains correlations for the off - resonant case . since asymptotic geometric discord is a positive function vanishing at @xmath138 and large values of @xmath131 , according to the rolle s theorem it must have a maximum at some intermediate detuning i.e. @xmath182 . in summary
, we have investigated the dynamical behavior of the geometric discord and negativity of a bipartite system including a two - level atom , prepared initially in a mixture of ground and excited states , and a quantised radiation field , prepared in a general pure number / coherent initial state .
the internal interaction is modelled by the jcm in the weak coupling approximation .
the evolution of the density matrix of the composite system is studied under pure dephasing process .
the results of this paper show that dephasing , induced by intrinsic decoherence , is competing with internal interaction to create an asymptotic value of quantum correlation after some coherence oscillation .
hence the amount of quantum correlation of the system reaches a stationary value , asymptotically .
the dynamical and asymptotic behavior of the geometric discord depend on the initial conditions and the system parameters .
indeed , the effects of dephasing can be amplified or weakened by adjusting both the parameters of the model and the initial conditions .
we have also shown that the asymptotic geometric discord reaches its maximum value at an intermediate non - zero value of atom - field detuning .
this system can be employed for implementation of the quantum remote state preparation protocol and the results help to improve the fidelity of this task .
the authors wish to thank the office of graduate studies of the university of isfahan for their support .
s. luo , phys .
a * 77 * , 042303 ( 2008 ) ; r. dillenschneider , phys . rev .
b * 78 * , 224413 ( 2008 ) ; m. s. sarandy , phys . rev . a * 80 * , 022108 ( 2009 ) ; m. ali , a. r. p. rau , and g. alber , phys . rev .
a * 81 * , 042105 ( 2010 ) ; g. adesso and a. datta , phys .
. lett . * 105 * , 030501 ( 2010 ) ; p. giorda , and m. g. a. paris , phys .
lett . * 105 * , 020503 ( 2010 ) . | in this paper , the dynamical behaviour of the geometric discord of a system consisting of a two - level atom interacting with a quantised radiation field described by the jaynes - cummings model has been studied .
the evolution of the system has been considered in the pure dephasing regime when the field is initially in a general pure state and the atom is initially in a mixed state .
dynamics of the geometric discord , as a measure of non - classical correlation , has been compared with the dynamics of negativity , as a measure of quantum entanglement . in particular , the influence of different parameters of system such as detuning and mixedness of the initial atomic state on the dynamics of geometric discord has been evaluated for when the field is initially in coherent and number states .
it is shown that for asymptotically large times , the steady state geometric discord of the system presents a non - zero optimum value at some intermediate value of detuning . |
the pupil provides a window into some of the processing that otherwise takes place invisibly inside the human brain .
hess and polt @xcite , @xcite as well as later kahneman and beatty @xcite , @xcite found evidence that linked emotional and cognitive processes to pupil dilations , and aston - jones et al .
@xcite , @xcite and joshi @xcite have provided a framework for understanding some of the anatomical processes that take place in regulating the gain of the networks involved , and why pupillary reactions are visible : at the core , the locus coeruleus - norepinephine ( lc - ne ) system operates in two different modes , _ tonic mode _ that regulates the overall level of preparedness or arousal and _ phasic mode _ that is involved in responding to task - relevant stimuli . as task difficulty increases , so will tonic mode activity , modulating the gain , which in turn leads to a increased performance and a stronger phasic response to task - relevant stimuli .
if , however , the arousal system and tonic activity mode increase beyond a certain peak point , the phasic responses decrease , leading to an explanation of the classical trade - off between arousal and optimal performance first analysed by yerkes and dodson @xcite . , where the largest phasic dilations are seen as compared to at [ a ] and [ c ] .
note that the graphs are not actual data to scale but is drawn for illustrative purposes .
( adapted from @xcite @xcite , resembling the classical yerkes - dodson relationship @xcite ) ] ' '' '' , highly focused task - specific attention [ b ] and distractible , scanning attention [ c ] .
the blue curve illustrate the level and fluctuations of the pupil size at each condition .
the baseline pupil size is shown in black , with the green area denoting the size of the response present when a task - relevant stimuli appear .
note that the drawing is not to scale . ] ' '' '' activity in lc - ne cells are further reflected in pupillary dilations @xcite , and the pupil can thus be interpreted as a marker of lc - ne activity ( see also fig . [
fig : phasictonic ] and fig .
[ fig : eyedilation ] ) .
baseline pupil size varies on a large scale of 3 - 4 mm as a response to changes in light levels @xcite , @xcite whereas variations caused by cognitive processes are much smaller , typically on the order of 0.5 mm or 15% compared to typical pupil sizes found in normal conditions @xcite .
the baseline pupil size is modulated by the tonic activity in lc - ne , and is never at rest ; it has been known for a long time to vary .
stark et al .
@xcite speculated that this could be part of an `` economical '' construction of the eye in the sense that there is no need for the eye to operate at a more narrow range , and in our previous study @xcite we also noted slow variations of the baseline pupil size of + /-10% on a timescale of 3060s .
task - evoked pupillary responses @xcite , @xcite ( tepr ) above the current baseline are caused by phasic activity in the lc - ne system , and by averaging over many similarly conditioned tests time - locked to the stimuli , other factors can be filtered out .
recent fmri studies by kuchinsky et al . have further established that activity in saliency networks triggered by attentional tasks are reflected in increased tonic pupil size , in contrast to the decreased pupil dilation typically observed when we are in a default resting state @xcite .
phasic activations of the lc - ne system in the noradrenergic ( ne ) neurons also play a role in rapid adaptation to changing conditions , as demonstrated by bouret et al .
@xcite , in that it may facilitate reorganisation of the innervated areas .
this allows for adaption of behaviour to changes in task conditions ; real or when they deviate from anticipation .
preuschoff et al .
@xcite have found that pupil dilations not only reflect decision making per se or the level of engagement , but also indicates surprise when committing mistakes in decision tasks , suggesting that ne plays a role in error signalling .
this appear similar to the negativity feedback components , which is an event related potential ( erp ) typically observed in eeg neuroimaging experiments 250 - 300 ms after participants realize that an incorrect choice was made @xcite .
while attention can be broadly understood as `` the appropriate allocation of processing resources to relevant stimuli '' @xcite , posner and petersen @xcite , @xcite , @xcite have shown that three systems , which regulate attention , are anatomically separate from other processing systems and carry out different cognitive roles as part of the attention networks .
these are : * alerting , * orienting and * executive control
. fan et al .
@xcite designed a behavioral experiment , known as the attention network test ( ant ) , to assess which of the network components are activated based on differences in reaction time when responding to visual cues .
we have in a previous experiment @xcite , measured task - evoked pupillar responses during the ant test in a longitudinal study of two subjects .
a stronger response was triggered by incongruent conditions in the conflict resolution decision task , likely involving the executive network .
this study expands the number of subjects , investigates the changes in mean pupil size over the experiment , and look at the relationship between the tonic level and the accuracy of the responses .
the procedure followed and the equipment used is identical to that described in @xcite , and is further illustrated in fig .
[ fig : ant - run - exp ] . in this present study , in total n=18 participants (
7 female and 11 male ) with a mean age of 25.3 years were tested once .
none used glasses or contact lenses , and all but one were right - handed .
the participant were all volunteers that were only allowed to complete the test if they gave consent to their data being used anonymously for research purposes .
the ant test itself is a standard paradigm in widespread use .
pupil size is recorded at 60 hz , and blink - affected periods are removed .
a hampel @xcite filter with a centered window of + /-83ms and a limit of @xmath0 is applied to remove outliers , and when data is not present in at least half the window , the center point is also removed .
this later part takes care of removing any samples immediately before and after blinks , to avoid accidental pupil size changes caused by distortion of the visible part of the eye .
finally data is downsampled to 100ms resolution with a windowed averaging filter . for the tepr calcuation ,
data is epoched to the cue presentation and individually scaled to the value at the start of the epoch . for the tonic pupil size
, a period of 1s immediately before target presentation is sampled to give a representative value without the phasic response , that in most conditions appear to fade away after around 2.5s after stimuli .
the pupil was further corrected for variations in head - distance by means of the eye - to - eye distance reported by the eye - tracker .
the mean pupil size was calculated in each of the 4 periods ( the initial trial round and the three actual blocks of reaction time tests ) as the average value of the filtered pupil data corrected for head - distance variations , which means it is representative of both the tonic pupil size and the ovelaid phasic responses .
' '' '' [ tab : rtant ] .average reaction- and attention network - times over all correct tests across all users ( @xmath1 sample standard deviation listed in parenthesis ) , in seconds .
[ cols="^,^,^,^",options="header " , ] left eye tonic pupil size , as measured immediately before target presentation , relative to each session s mean , and the reaction times , are listed across all subjects of the present study , and for both participant a and b over all sessions of the longitudinal study , for incongruent conditions , divided into groups of correct and incorrect responses .
the mean reaction time ( @xmath2 ) differ between correct and incorrect responses in a significant way ( welch t - test @xmath3 , @xmath4 and @xmath5 respectively , p<0.000001 ) for both a and b. the means of the tonic pupil size ( psz ) differ significantly between correct and incorrect responses for a ( welch t - test @xmath6 , @xmath7 ) ; for b and the participants of the present study , the means between the conditions do not show a statistically significant difference .
almost identical results are found for right eye pupil sizes ( not listed here ) . [
tab : ok_vs_nok_incongruent ] ) are 0.345 and 0.301 respectively .
initiation of each of the 3 rounds of the session are marked with dashed lines .
an initial increased pupil dilation diminishes over time as entraining takes place , with a slight increase towards the end .
it can also be seen that , in this case , each round starts out with an increase pupil dilation . ]
[ fig : tonic_pupil_20160112150851 ] shows an illustrative sample of how the mean pupil size ( corrected for variations in head distance ) varies over the course of the initial training round and the three actual trial blocks .
left and right pupil size are slightly different for this particular subject , but there is good correlation between variations of the two ( pearson s @xmath8 ) . an regression corresponding to a low pass filter ( a 2nd order polynomial fit ) is shown overlaid , and can explain approximately 30 - 35% of variance ( explained variance @xmath9 and @xmath10 respectively ) .
it also appears as if each block has a slightly larger tonic pupil size initially followed by a decline of approximately 10% .
the means for each block also apper to differ , with the initial training round having the larger tonic pupil size . , @xmath11 , @xmath12 , all with a confidence level @xmath13 ) .
the differences between the other blocks are not statistically significant . ] , @xmath14 , @xmath15 , all with a confidence level @xmath13 ) .
the differences between the other blocks are not statistically significant .
however , for a there are no statistically significant differences between the blocks ; the variations between the 4 blocks are comparatively much smaller than than what is seen for other participants . ] when comparing the mean tonic pupil size between the initial training round and the three actual trial blocks , there are statistically significant differences across all participants of the present study , and also for participant b of the longitudinal study . for participant a , however , there are no statistically significant differences .
see fig .
[ fig : qmeans_all ] and fig .
[ fig : qmeans_ab ] average fixation density maps , adjusted for accidental mis - calibrations , were built for each experiment , and were compared between conditions .
we did , as expected , see recognisable differences when the target presentation was above vs below the fixation cross , but we were not able to detect any significant spatial differences between congruency conditions nor between cue conditions .
the results of this study , with a larger population , supports our previous findings : there is a difference in the incongruent vs congruent / neutral flanker scenarios in that an incongruent condition solicit a larger pupillary response compare to the other two conditions .
as the age group is different compared to the previous study , there are indications that the results may be robust and can translate to different settings .. in most cases we see a high correlation ( r values from 0.80.95 ) between left and right pupil size , although a few have what may be less than optimal tracking .
we can not conclude any significant difference in the pupil dilation responses between the two eyes , but we notice that the significance level of the difference between the incongruent and the neutral condition is higher and lasts slightly longer for the right eye .
in addition , we also found a significantly different response when subjects replied incorrectly , which happens much more frequently for the inconguent condition .
this response may be related to the adaptation and required reorganization reported by bouret et al .
@xcite and/or to the surprise elements reported by preuschoff et al .
@xcite .
thus , the phasic response reported here as well as in our previous study can be divided into two components that apparently cause a higher level of lc - ne activations : one related to the incongruent condition and one to the incorrect reply . comparing the mean relative pupil size over the 4 parts of the experiment ( training round and 3 blocks of tests ) we found that for the subjects of the present study , as well as for subject b of the longitudinal study , the training round had a statistically significant higher level , around 5% , compared to the other three blocks that averaged around -2% .
the subject a of the longitudinal study , however , did not show any such variation between the blocks .
we hypothesize that this may point to differences in individual characteristics , behavoiur or preferences .
further , comparing relative normalized tonic pupil sizes ( excluding the phasic responses ) showed a statistically significant difference between the level immediately before an incorrect reply compared to the level before a correct reply for subject a of the longitudinal study but not for subject b nor for the participants of the present study .
however , while for subject b the levels are almost identical , there is a larger difference even if not statistically significant for the participants of this study , and it therefore can not be ruled out that participants could fall in different groups that , with more data , would reveal more individual variation .
we also point out that the possible familiarity effects of higher pupillary responses mainly in the two first complete experiments were not tested for in the present experiment , since it was performed only once for each participant .
we do , however , see hints at an overall adaptation , as the average ( tonic ) level decreases as initial entraining to the tasks take place , with a flat or in some cases slightly increased tonic levels towards task completion .
this appear similar to the _ familiarity effect _ reported by hyn et al .
@xcite we were not able to find any spatial differences in eye movements , at the resolution we worked with , that was related to the conditions of the test , apart from the up - down position of the target .
this work is supported in part by the innovation fund denmark through the project eye tracking for mobile devices .
hess eh , polt jm . .
available from : http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=retrieve{&}db=pubmed{&}dopt=citation{&}list{_}uids=14401489[http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=retrieve\{&}db=pubmed\{&}dopt=citation\{&}list\{_}uids=14401489 ] .
the british journal of ophthalmology .
available from : http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1039657{&}tool=pmcentrez{&}rendertype=abstract[http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1039657\{&}tool=pmcentrez\{&}rendertype=abstract ] .
bkgaard p , petersen mk , larsen je . in : antona m , stephanidis c , editors . assessing levels of attention using low cost eye tracking .
cham : springer international publishing ; 2016 .
p. 409420 .
available from : http://dx.doi.org/10.1007/978-3-319-40250-5{_}39[http://dx.doi.org/10.1007/978-3-319-40250-5\{_}39 ] .
kuchinsky se , panda nb , haarmann hj . in : schmorrow dd , fidopiastis mc , editors . linking indices of tonic alertness :
resting - state pupil dilation and cingulo - opercular neural activity .
cham : springer international publishing ; 2016 .
p. 218230 .
available from : http://dx.doi.org/10.1007/978-3-319-39955-3{_}21[http://dx.doi.org/10.1007/978-3-319-39955-3\{_}21 ] .
posner mi , petersen se . annual review of neuroscience .
available from : http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3413263{&}tool=pmcentrez{&}rendertype=abstract$\backslash$nhttp://www.ncbi.nlm.nih.gov/pubmed/2183676[http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3413263\{&}tool=pmcentrez\{&}rendertype=abstract$\backslash$nhttp://www.ncbi.nlm.nih.gov/pubmed/2183676 ] . | cognitive processes involved in both allocation of attention during decision making as well as surprise when making mistakes trigger release of the neurotransmitter norepinephrine , which has been shown to be correlated with an increase in pupil dilation , in turn reflecting raised levels of arousal . extending earlier experiments based on the attention network test ( ant ) , separating the neural components of alertness and spatial re - orientation from the attention involved in more demanding conflict resolution tasks , we demonstrate that these signatures of attention are so robust that they may be retrieved even when applying low cost eye tracking in an everyday mobile computing context .
furthermore we find that the reaction of surprise elicited when committing mistakes in a decision task , which in the neuroimaging eeg literature have been referred to as a negativity feedback error correction signal , may likewise be retrieved solely based on an increase in pupil dilation . |
back to 1950 , the paper by matthews and salam @xcite and the subsequent works by salam in 1952 @xcite clarified the quantum field - theoretical approach to the electrodynamics of ( massive ) charged scalar particles . ever since , the problem of extending this investigation to include the case of ( massive ) charged vector particles became mandatory in view of the theoretical evidence for the role of charged and neutral spin-1 particles that should couple to the charged and neutral currents . from 1960 ,
komar and salam @xcite , salam @xcite , lee and yang @xcite and delbourgo and salam @xcite , gave a highly remarkable push for the understanding and construction of a fundamental theory for the microscopic interaction between charged vector bosons and photons .
further works by tzou @xcite , aronson @xcite and velo and zwanziger @xcite summed up efforts to the previous papers and the final conclusion was that a consistent unitary and renormalizable quantum field - theoretical model would be possible only in a non - abelian scenario with a higgs sector that spontaneously breaks the gauge symmetry to give mass to the vector bosons , without spoiling the unitarity bounds for the cross section of the scattering of longitudinally polarized charged vector bosons @xcite . here
, we adopt this framework : the electrodynamics of ( massive ) charged vector bosons with a non - minimal dipole - type term coupling to the gauge field . in the case of the massless photon , this ensures a @xmath0value for the g - ratio of spin-1 particles . we should however point out that this dipole - type interaction is non - minimal from the viewpoint of an abelian symmetry ; if we take into account the local @xmath1 symmetry that backs the plus- and minus - charged vector bosons , the dipole coupling in the action is actually a minimal ( non - abelian , @xmath1 ) interaction term after spontaneous symmetry breaking has taken place . in the present paper ,
our main effort consists in pursuing a detailed investigation of the semi - classical aspects of the charge and spin interactions for massive charged matter of a vector nature .
our central purpose is to compare the features and specific profiles of the influence of the spin of the charge carriers on interaction potentials ( electromagnetic or a more general @xmath2 interaction ) between two different categories of sources / currents : fermionic and spin-1 bosonic .
we shall present the conclusions of our comparative procedure at the end of our calculations . at this point
, we highlight that the literature in the topic contemplates a great deal of articles discussing the structure of the electromagnetic current and electrodynamical aspects of spin-1 charged matter @xcite-@xcite .
we pursue an investigation of an issue not considered in connection with spin-1 charged matter : the spin- and velocity - dependence of the interaction potential associated with ( pseudo-)scalar sources and pseudo - vector currents that interact by exchanging scalar and vector mediators , respectively . for the spin-1
, these specific cases have not been addressed to in the literature .
these extra sources / currents may not necessarily be associated to the electromagnetic interaction in that they do not follow from the @xmath2 symmetry of the electromagnetism .
we may be describing a new force between these extra sources / current whose origin could be traced back to some more fundamental physics .
the case of the usual vector current is reassessed here and our results match with the ones in the literature .
it is worthy mentioning that fermionic sources / currents can display a wide range of spin- and velocity - dependent interparticle potentials @xcite@xcite@xcite , and several of them have been reconsidered in this article .
the electrodynamics of ordinary fermionic matter is very well understood , from the macroscopic to the quantum level .
scalar and vector bosonic charged matter , on the other hand , experience a richer variety of interactions when coupled to the electromagnetic field .
so far , most of the theoretical and experimental literature dealing with macroscopic interactions consider only spin-1/2 matter , i.e. , fermionic sources / currents @xcite-@xcite .
this preference is naturally due to the fact that ordinary charges in matter are carried by electrons .
elementary spin-1 charged particles are difficult to observe , since the only known examples are the @xmath3 gauge bosons , whose mass is too large - and lifetime too short - to allow direct inspection ( its full width is @xmath4 @xcite ) .
though not stable enough to be directly handled , elementary spin-1 particles have , in principle , their own electrodynamics and it is of theoretical interest to study how it deviates , or not , from its fermionic counterpart .
on the other hand , at the atomic and nuclear level , it would be a good motivation to have a spin- and velocity - dependent expression for the electromagnetic potential between ionised spin-1 ( charge @xmath5 ) atoms and charged spin-1 nuclei or hypernuclei @xcite@xcite .
we highlight here a particular feature of bosonic charge carriers as far as the electromagnetic interactions are concerned . from a purely macroscopic point of view
, the maxwell equations address the problem of determining field configurations from given charge and current densities and a number of duly specified boundary conditions .
they do not take into account the microscopic nature of these sources ( @xmath6 and @xmath7 , respectively ) . if a microscopic description of charged matter in terms of classical fields is given (
based on a local @xmath2 symmetry ) , the particular aspects of the charge carriers become salient and london - type terms @xcite may arise . we shall be more specific about this point at the final section of our paper , where we render more evident the peculiarities of the spin of the charge carriers in the electromagnetic and general abelian interactions .
our paper is outlined as follows : in section ii , we introduce the concept of potential and briefly discuss the kinematics involved , as well as present the notations and conventions employed . in section iii , we calculate the sources / currents and potentials , discussing the similarities between spin-1/2 and spin-1 matter .
we consider , in section iv , another possible field representation for the spin-1 charged carrier , namely via a rank-2 tensor field , and we also work out the corresponding source / current@xmath8source / current interaction potentials .
we present our conclusions and perspectives in section v. natural units are adopted throughout , where @xmath9 , and we assume @xmath10 .
since we are interested in comparing the low - energy behavior of spin-1/2 and spin-1 as the interacting matter sector , it is convenient to work in the non - relativistic ( nr ) limit of the interactions . for simplicity
we choose to work in the center of mass ( cm ) reference frame , in which particle 1 has initial and final 3-momenta given by @xmath11 and @xmath12 , respectively . here
, @xmath13 is the average momentum of particle 1 before and after interaction , while @xmath14 is the momentum transfer carried by the intermediate boson - see fig .. momenta assignments for the tree - level interaction between sources / currents 1 and 2.,scaledwidth=50.0% ] the potential is calculated through the first born approximation @xcite , i.e. the fourier transform of the amplitude , @xmath15 , with respect to the momentum transfer @xmath16 where we explicitly assume that the amplitude contains terms dependent on the velocity of particle 1 .
it is clear from the nr approximations of the sources / currents presented below that not only velocity - dependent terms will arise in the potentials , but also spin - dependent ones .
this is due to our choice of keeping terms in the currents which go beyond the zeroth order , thus static , case @xcite . in what follows
, we shall consider only elastic interactions , which , together with energy conservation , imply @xmath17 and @xmath18 , pointing out that @xmath19 is a space - like 4-vector : @xmath20 .
furthermore , in the amplitude we will keep only terms up to second order in @xmath21 .
the feynman rules for the tree - level diagram above are equivalent to taking @xmath22 as the interaction vertices , where @xmath23 are the matter sources / currents associated with particles 1 and 2 , respectively .
the corresponding amplitude may then be written as @xmath24 where @xmath25 is the momentum - space propagator of the intermediate boson with adequate lorentz indices as the case may be . in this paper , we are interested in interactions mediated by massive neutral scalars ( klein - gordon - type ) and spin-1 vector particles ( proca - type ) , whose momentum - space propagators are : @xmath26 and @xmath27 the sources / currents @xmath28 and @xmath29 , as representatives of the respective vertices , must translate the different possible couplings to the gauge sector : scalar , pseudo - scalar , vector and pseudo - vector .
the vector currents are obtained from the lagrangian through the noether method , but , for the bosonic case , we consider only first order in the electromagnetic coupling constant , @xmath30 , since we wish to compute @xmath31 to second order in @xmath30 .
we shall discuss this in more details in subsection b. the other sources / currents are built based on the principle that they should be bilinear in the field and its complex conjugate and reflect the desired symmetry .
below , we recall some basic properties from spin-1/2 and spin-1 starting from the rest frame where the essential degrees of freedom become apparent , and then lorentz - boosting it to the lab frame
. this will be important when we calculate the nr limit of each source / current and then apply them to extract the interparticle potential . here
, we choose to work the spin-1 in its vector field representation , but , in section iv , we shall discuss it in its less usual tensor field representation @xcite@xcite@xcite . for the sake of completeness , we present the well - known properties of dirac fermions .
the dirac equation for the positive - energy spinors is , in momentum - space , @xmath32 u(p ) = 0 \label{diraceq}\ ] ] with the corresponding equation for the spinor conjugate , @xmath33 , with the attribution @xmath34 . by using the gamma matrices in the dirac representation
, the positive - energy spinor can be explicitly written in the nr limit as @xmath35 with @xmath36 are the basic spinors satisfying @xmath37 and @xmath38 denotes a normalization constant . manipulating eq . and
its conjugate , we obtain the gordon decomposition of the vector current , @xmath39 expressed in terms of the more convenient variables @xmath40 and @xmath41 , where we defined the spin matrix as @xmath42 $ ] .
similar decompositions are also possible for other types of bilinear forms : for example , using eq . and
its conjugate , we show that @xmath43 the gordon decomposition of the vector current associated with the noether theorem , eq . , yields a first term proportional to the density of the field ( @xmath44 ) , while the second one presents a coupling between momentum transfer , @xmath19 , and the spin matrix , @xmath45 , of the respective representation .
this connection will be responsible for the zeroth order magnetic dipole moment interaction of the lepton with gyromagnetic ratio @xmath46 @xcite .
interestingly enough , this simple structure found for the gordon decomposition for the spin-1/2 will have a perfect analog in the spin-1 case , if we correctly include a non - minimal coupling - see subsection b. our interparticle potentials shall be derived from the ( effective ) lagrangian describing the interactions between the exchanged ( scalar or vector ) particles and the sources / currents of the fermionic particle .
it reads as follows below : @xmath47\psi , \label{lag_g_ferm}\ ] ] where the different couplings are all dimensionless .
the scalar ( s ) and pseudo - scalar ( ps ) sources are not conserved and , since we are dealing with a massive dirac field , we also have a non - conserved pseudo - vector ( pv ) current ( see eq . ) .
as pointed out by boulware @xcite , the proca - type field , whenever coupled to a conserved current , leads to a renormalizable model .
for this reason , we shall consider the coupling between the @xmath48 current and proca - type field as an effective interaction .
the uv divergence does not harm our purposes here , since we are interested in the low - energy regime described by the interparticle potential .
actually , our external ( on - shell ) sources / currents are tailored in the nr regime , meaning that we are working much below the scale where uv divergences show up .
henceforth we shall denote the coupling constant @xmath49 whenever considering the electromagnetic interaction . in the previous subsection
, we have revised some basic aspects concerning spin-1/2 particles . here
, we would like to follow up with a similar discussion for massive charged spin-1 particles in order to comparatively study how differently ( or not ) spin-1/2 and spin-1 sources / currents interact in the nr limit .
we start off discussing the vector field representation for a massive charged spin-1 particle , whose dynamics can be obtained from the following lagrangian : @xmath50 where @xmath51 is the mass and @xmath52 the gauge - covariant field strength , given by @xmath53 , with @xmath54 .
the last term in eq . is a non - minimal coupling between the bosonic matter fields and the field - strength of the interaction mediator and its importance lies in the fact that it asserts that @xmath46 in tree - level for the spin-1 particle @xcite@xcite@xcite ( see also references therein ) .
the field @xmath55 has , in principle , four degrees of freedom and these are reduced through the constraints introduced by the equations of motion , which read @xmath56 with analogue ones for the complex conjugated field .
our goal is to characterize the source / current@xmath8 source / current interaction and for this , at tree - level , the aforementioned free source / current are to be described by bilinears in the spin-1 matter fields , where the derivatives are the ordinary ones . in other words , since the asymptotic states are the particle states composed only of @xmath57 fields ( we are now taking the free field equations ) , we have the subsidiary condition @xmath58 .
this constraint is better seen in momentum - space , where we have @xmath59 , so that @xmath60 where , in order to avoid a cumbersome notation , we have used the same symbol to denote the field in position- and momentum - space , as there is no risk of confusion .
this equation relates different components of the vector field and shows that the time - component is proportional to the longitudinal projection of @xmath61 and that @xmath62 is first - order in velocity , playing the role of a `` small '' component similar to the spin-1/2 case .
it also allows us to write @xmath63 in its rest frame as @xmath64 , where @xmath65 stands for the dimensionless polarization 3-vector in the rest frame of the free particle and @xmath66 is a normalization constant . the next step is to bring the system from rest into motion ( i.e. , to the lab frame ) via an appropriate lorentz transformation . in doing so
, we obtain @xmath67 \label{wlab}\ ] ] and with this , two comments are in order : * eq . tells us that , in the rest frame ( where @xmath68 ) , all the information about the vector field is contained in its spatial part . in this frame
the only 3-vector available is the polarization , i.e. , its spin , which again enforces the vector character of the field . *
the normalization constant is given by @xmath69 .
we wish to make some considerations on the vector currents , both global and local , associated to the lagrangian . at the global stage ,
i.e. , prior to the gauging of the @xmath2 symmetry , the noether current ( in configuration - space ) reads @xmath70 where no covariant derivative is involved , so @xmath71 . upon the gauging of the symmetry , and by including the non - minimal coupling g - term , the current changes into @xmath72 now , an important remark concerning current conservation : to calculate the current - current potential at the tree - level approximation ( one - boson exchange ) , as we are doing here , we actually only consider the current to the first order in @xmath30 .
this amounts to neglecting the terms in the gauge potential present in the current in eq .. in other words : to calculate the potential to order @xmath73 , we consider the current only up to order @xmath30 .
therefore , to the desired order in the coupling constant , the vector current is the globally conserved one , plus the non - minimal @xmath74-term . at this perturbative level ,
the current @xmath75 is given by the global and non - minimal terms , as displayed in the second line of eq .
below : @xmath76\phi + \nonumber \\ & + & ig_{v}\left [ w^{\ast \ , \mu\nu}w_{\nu } - w^{\mu\nu}w^{\ast}_{\nu } + ( g-1 ) \partial_{\nu}\left ( w^\nu w^{\ast \mu } - w^{\ast \nu } w^\mu \right ) \right]a_{\mu } + \nonumber \\ & + & ig_{pv } \left [ \tilde{w}^{\ast \ , \mu\nu}w_{\nu } - \tilde{w}^{\mu\nu}w^{\ast}_{\nu } \right]a_{\mu } , \label{lag_g_bos}\end{aligned}\ ] ] where the coupling constants have the following canonical mass dimensions : @xmath77 = - [ g_{ps } ] = 1 $ ] and @xmath78 = [ g_{pv } ] = 0 $ ] , and the interactions between the sources / currents and the mediating ( scalar or vector ) bosons are all displayed . by using the equations of motion and the ( free ) subsidiary conditions , we may re - write the vector current in momentum space as @xmath79 in which @xmath80 stands for the spin matrix in the @xmath81-representation of the lorentz group and , as for the spin=1/2 case , we have set @xmath82 . in the equation above we left the g - factor explicit but in the nr limits we shall use its tree - level value , @xmath46 .
eq . is nothing but the gordon decomposition for the vector field representation of a massive and charged spin-1 and the similarity with the spin-1/2 case , eq .
, is remarkable .
this gordon decomposition for spin-1 particles was found independently and agrees with the results of ref.@xcite .
we denote the dual of @xmath83 , as given in the @xmath84 and @xmath48 couplings in eq .
, by means of @xmath85 .
we also emphasize that the ( global ) @xmath48 bosonic current is conserved in a topological sense , i.e. , without making use of either the equations of motion or the symmetries of the lagrangian . as in the case of spin-1/2 ,
the @xmath86 and @xmath84 sources for spin-1 are equally not conserved . in the next section ,
we discuss the momentum - space sources / currents for spin-1/2 and spin-1 in the nr limit and calculate the interparticle potentials generated by the exchange of scalar and vector mediating bosons .
following the indications above , we present below the nr limit of the sources / currents , which may be extracted ( in configuration - space ) from lagrangians and as @xmath87 or @xmath88 , as the case may be . in the following ,
the fields are already normalized with @xmath89 and @xmath90 . from now on , for the sake of clarity , we shall adopt the variables @xmath91 instead of @xmath92 , so @xmath93 and @xmath94 ; similar definitions hold for @xmath95 and @xmath96 .
we point out that the matter sources / currents developed in this section are for particle 1 , i.e. , the left vertex in fig .. the second source / current may be obtained by taking @xmath97 , @xmath98 in the first one . at this point
it is convenient to reinforce the difference between polarization and spin of the particle . in the fermionic case
, we shall denote the expectation value of its spin , @xmath99 , by contracting the basic spinors @xmath36 with the pauli matrices , namely , @xmath100 . the same idea is applied to the bosonic case , where the spin matrix of the particle is @xmath101 , with @xmath102 , and its expectation value is given by @xmath103 . a remark about our notation : in the process of calculating the amplitude factors , the form @xmath104 and @xmath105 , with @xmath106 labeling the particle , will appear frequently .
these indicate a possible spin - flip between initial and final states , and we shall leave the @xmath107 s explicit in the final expressions , despite the fact that , in general , low - energy interactions will not induce a change in the spin orientation of the particles involved . by means of the relations presented in the previous sections , we can carry out the nr expansion of the @xmath86 and @xmath84 of the spin-1 sources . for the @xmath86source of spin-1/2 particles , we can use the dirac spinor , eq .
, and its conjugate , to obtain @xmath108 - \frac{i}{4m_1^{2}}\vec{q}\cdot \left ( \vec{p } \times \langle \vec{\sigma } \rangle_1
\right ) \biggr\ } \label{us}.\end{aligned}\ ] ] for the @xmath86source of the spin-1 particles , we obtain the expression below : @xmath109 \right\},\end{aligned}\ ] ] which can be written in a more enlightening way if we note that the polarization and momenta 3-vectors satisfy @xmath110 .
this allows us to set the spin-1 scalar source as @xmath111 \right\ } \label{ws}.\end{aligned}\ ] ] by proceeding analogously , we can find the pseudo - scalar sources listed below @xmath112 and here we note that both pseudo - scalar sources have the same functional form characterized by the coupling between spin and momentum transfer . in the following , we shall present the potentials .
let us calculate a particular case , namely that of @xmath113 interaction , for both @xmath114 and @xmath115 sources .
we start off with the fermionic case .
as discussed in the methodology , the general amplitude is given by eq . and , in this case , @xmath116 where we have used the klein - gordon propagator , eq . , and the momentum assignments of fig .
: @xmath117 at this point , we use the fermionic scalar source presented in eq .. this scalar source remains unchanged by taking @xmath118 and @xmath119 , since it is quadratic in the momenta ( it will only be necessary a change in labels : @xmath120 ) . if we keep terms up to second order in the momenta , the amplitude becomes @xmath121 + \nonumber \\ & - & \frac{i}{4 } \
, \left ( \vec{q } \times \vec{p } \right ) \cdot \left [ \frac{\delta_1 \langle \vec{\sigma } \rangle_2}{m_2 ^ 2 } + \frac{\delta_2 \langle \vec{\sigma } \rangle_1}{m_1 ^ 2 } \right ] \biggr\}\end{aligned}\ ] ] and the fourier transformation given in eq .
yields the potential : @xmath122 + \nonumber \\ & + & \frac{\left ( 1 + m_\phi r \right)}{4r^2 } \vec{l } \cdot \left ( \frac{\delta_1 \langle \vec{\sigma } \rangle_2}{m_2 ^ 2 } + \frac{\delta_2 \langle \vec{\sigma } \rangle_1}{m_1 ^ 2 } \right ) \biggr\}. \label{s_s_ferm } \end{aligned}\ ] ] as expected , the dominant term is the monopole - monopole given by the yukawa interaction , @xmath123 .
we also obtain second - order corrections to this term in the form of a monopole - dipole with spin - orbit couplings .
the amplitude of the bosonic @xmath113 case presents the same structure as given in eq . , but with the sources below : @xmath124 we follow the same program , i.e. , by using the nr limit of the bosonic scalar source , eq .
, we obtain @xmath125 + 1 \leftrightarrow 2 \right ] \biggr\ } \nonumber\end{aligned}\ ] ] and , performing the fourier integral , we get @xmath126 + 1 \leftrightarrow 2 \right ] \biggl\ } , \label{s_s_rep_vet}\end{aligned}\ ] ] where the parameter @xmath127 in the last term may assume the values @xmath128 ; in eq .
its value is @xmath129 .
however , in the particular situation described in section iv ( a 2-form description of the massive spin-1 particle ) , the potential also appears , but with @xmath130 .
here we notice that this potential describes the same interactions as in the fermionic case , namely , the yukawa factor and a spin - orbit term , but it also displays a polarization - polarization " interaction term
. this observation will be present in other situations in our comparison between the fermionic and bosonic potentials .
below , we quote the results for the other bosonic potentials : @xmath131 the fermionic @xmath132 and @xmath133 potentials have the same functional form as the bosonic ones and , if we neglect a contact term ( that is , a @xmath134 term ) , these results are very similar to those obtained by moody and wilczek @xcite in the context of fermionic sources exchanging axions .
indeed , they only differ by a mass factor due the canonical mass dimension of our coupling constants .
firstly , let us focus on the bosonic case , more precisely , on the spin-1 vector current , @xmath135 .
for the @xmath136 component in eq . , we have , in momentum - space @xmath137 + \nonumber\\ & + & ( g-1)\frac{e_1}{2m_1 ^ 2 } \left [ ( \vec{p } \cdot \vec{\epsilon_1^{\ , \ast}})(\vec{q } \cdot \vec{\epsilon}_1 ) - ( \vec{q } \cdot \vec{\epsilon_1^{\ , \ast}})(\vec{p } \cdot \vec{\epsilon}_1 ) + ( \vec{q } \cdot \vec{\epsilon_1^{\ , \ast}})(\vec{q } \cdot \vec{\epsilon}_1 ) \right],\end{aligned}\ ] ] where the entire second line comes from the @xmath74-factor correction term .
if @xmath138 , that line vanishes and we would have an incomplete description of the electromagnetic interaction experienced by the vector boson , as seen from eq .. rearranging the equation above and taking the correct @xmath139value , we obtain the time - component of the spin-1 vector current in the form @xmath140 , \label{wv0}\ ] ] while , by taking @xmath141 in eq . , we get the spatial component as given by @xmath142 , \label{wvi}\ ] ] which is very similar with its spin-1/2 counterpart , as we shall see in the following .
the spin-1 pseudo - vector current is simpler : according to its expression in eq .
, we can obtain the following nr limits for @xmath136 and @xmath143 as : @xmath144 \left(\vec{q } \times \vec{\epsilon_1^{\ , \ast}}\right)_{i } - \left [ \left(\vec{p}+\frac{1}{2}\vec{q}\right ) \cdot \vec{\epsilon_1^{\ , \ast } } \right]\left(\vec{q } \times \vec{\epsilon}_1 \right)_{i } \biggr\}. \label{wpvi}\end{aligned}\ ] ] we present below the corresponding fermionic currents with the purely vector cases also carrying the electromagnetic coupling constant : @xmath145 + \frac{i}{4m_1^{2 } } \left ( \vec{q } \times \vec{p } \right)\cdot \langle \vec{\sigma } \rangle_1 \biggr\ } \label{uv0 } \\ & \ , & \nonumber \\
j^{i}_{v(s=1/2 ) } & = & e_1 \left [ \frac{\delta_1}{m_1}\vec{p}_{i } - \frac{i}{2m_1 } \epsilon_{ijk}\vec{q}_{j}\langle \vec{\sigma}_{k } \rangle_1 \right ] , \label{uvi } \\ & \ , & \nonumber \\
j^{0}_{pv(s=1/2 ) } & = & \frac{g^1_{ps}}{m_1}\vec{p}\cdot \langle \vec{\sigma } \rangle_1 , \label{upv0 } \\ & \ , & \nonumber \\ j^{i}_{pv(s=1/2 ) } & = & g^1_{pv } \biggl\ { \left [ 1 - \frac{1}{4m_1^{2}}\left ( \vec{p}^{\ , 2 } - \frac{1}{4}\vec{q}^{\ , 2 } \right ) \right]\langle \vec{\sigma}_{i } \rangle_1 - \frac{i\delta_1}{4m_1^{2 } } \left ( \vec{q } \times \vec{p } \right)_{i } + \nonumber \\ & + & \frac{1}{2m_1^{2 } } \left [ \left ( \vec{p}\cdot \langle \vec{\sigma } \rangle_1 \right)\vec{p}_{i } - \frac{1}{4}\left ( \vec{q}\cdot \langle \vec{\sigma } \rangle_1 \right)\vec{q}_{i } \right ] \biggr\}. \label{upvi } \end{aligned}\ ] ] unlike the @xmath146 and @xmath48 currents of the spin-1/2 case , the @xmath146 and @xmath48 currents of the spin-1 field , as defined in lagrangian ( marked by the @xmath147 and @xmath148 coupling constants ) , always carry an explicit derivative . at high energies , this extra momentum factor must dominate , but not in the nr limit .
this extra derivative affects the way these currents interact with the gauge sector , specially in the case of spin-1 .
furthermore , it is interesting to emphasize other differences between the fermionic and bosonic currents above . by comparing the vector currents ,
we notice essentially an extra term in the bosonic case associated with the contribution of the polarization , @xmath149 .
the @xmath48currents display more remarkable differences .
for example , for the @xmath136 component , eqs . and , we observe that the spin of the fermion couples to the average momentum , @xmath13 , while for the bosons the spin couples to the momentum transfer , @xmath14 . for the @xmath143 components , we have many differences due to the spin terms ( see eqs . and ) .
these special features will be responsible for the different behaviors of the bosonic and fermionic potentials .
let us now discuss the potentials involving @xmath146 and @xmath48 currents mediated by a proca - type particle .
as already done in the previous subsection , we shall exemplify the calculation through a particular configuration , and then quote the final results . we choose to calculate one of the simplest cases , the @xmath150 interaction between @xmath114 currents , since this case contains all the necessary elements to understand the other ( lengthier ) evaluations .
the amplitude is given by @xmath151 where we have used current conservation , @xmath152 , to eliminate the longitudinal contribution of the proca - type propagator , eq .. in this particular case @xmath153 , both currents are conserved .
we need therefore to simplify the contraction @xmath154 and , according to our assumptions , only the @xmath155 piece will contribute in the nr limit ; so , @xmath156 and we finally calculate the fourier integral to obtain @xmath157 in a similar way , the case with @xmath115 currents leads to @xmath158 + \frac{(1+m_a r)}{2m_1 r } [ \langle \vec{\sigma } \rangle_1 \times \langle \vec{\sigma } \rangle_2 ] \cdot \hat{r } \biggl\ } \label{v_pv_ferm}\ ] ] and here the @xmath48 current is not conserved ; however , the conservation of the @xmath146 current ensures that @xmath159 term of the proca propagator drops out .
as we have already seen in the discussion of the currents , the @xmath48 currents of @xmath115 and @xmath114 exhibit remarkable differences and these explain why the potentials behave so differently .
we anticipate that only this fermionic case yields the interaction with the @xmath160 dependence and the spin - spin contribution of the form @xmath161 . in the bosonic case
, we have the @xmath162 dependence , but this is not exclusive of the @xmath150 interaction , it also appears in the @xmath133 bosonic potential .
the other potentials , @xmath163 and @xmath164 , are listed below .
first , the fermionic ones : @xmath165 + \nonumber \\ & - & \frac{(1+m_a r)}{2 r^2 } \vec{l } \cdot \biggl [ \left ( \frac{\delta_1 \langle \vec{\sigma } \rangle_2}{2m_2 ^ 2 } + \frac{\delta_2 \langle \vec{\sigma } \rangle_1}{2m_1 ^ 2 } \right ) + \left
( \frac{\delta_1 \langle \vec{\sigma } \rangle_2}{m_1 m_2 } + \frac{\delta_2 \langle \vec{\sigma } \rangle_1}{m_1 m_2 } \right)\biggl ] + \nonumber \\ & + & \frac{1}{4m_1 m_2 r^{2 } } \langle \vec{\sigma } \rangle_1 \cdot \langle \vec{\sigma } \rangle_2 \left ( 1 + m_{a}r + m_{a}^{2}r^{2 } \right ) + \nonumber \\ & - & \frac{1}{4m_1 m_2 r^{2 } } \left ( \langle \vec{\sigma } \rangle_1 \cdot \hat{r } \right ) \left ( \langle \vec{\sigma } \rangle_2 \cdot \hat{r } \right ) \left(3 + 3m_{a}r + m_{a}^{2}r^{2 } \right)\biggl\ } \ , , \label{v_v_ferm } \end{aligned}\ ] ] where we can read the electromagnetic potential by setting @xmath166 ; here @xmath167 and @xmath168 stand for the respective ( electric ) charges of the particles . for the @xmath164 case we obtained @xmath169 + \nonumber
\\ & + & \left ( \langle \vec{\sigma } \rangle_1 \cdot \vec{p } \right ) \left ( \langle \vec{\sigma } \rangle_2 \cdot \vec{p } \right ) \left ( \frac{1}{m_1 m_2 } + \frac{1}{2m_1 ^ 2 } + \frac{1}{2m_2 ^ 2 } \right ) + \nonumber \\ & + & \frac{\left(1+m_a r \right)}{4 r^2 } \vec{l } \cdot \left ( \frac{\delta_2 \langle \vec{\sigma } \rangle_1}{m_2 ^ 2 } + \frac{\delta_1 \langle \vec{\sigma } \rangle_2}{m_1 ^ 2 } \right ) + \nonumber \\ & + & \frac{1}{8r^2 } \left ( \langle \vec{\sigma } \rangle_1 \cdot \hat{r } \right ) \left ( \langle \vec{\sigma } \rangle_2 \cdot \hat{r } \right ) \left ( \frac{1}{m_1 ^ 2 } + \frac{1}{m_2 ^ 2 } \right ) \left ( 3 + 3m_{a}r + m_{a}^{2}r^{2 } \right)\biggl\ } + v^{s=1/2}_{pv - pv - long } \ , , \label{pv_pv_ferm}\end{aligned}\ ] ] where @xmath170 the fermionic @xmath164 potential presents some peculiar aspects as an inheritance of the non - conserved @xmath48 currents .
first of all , in contrast to other cases , we had to take into account the longitudinal part of the proca propagator in our calculations . by doing that
, we obtain the contribution we have denoted by @xmath171 in eq .. for this reason , this potential is not well defined in the massless limit @xmath172 of the exchanged particle and the excitation of the unphysical spin-0 component of the massive vector boson ( @xmath173 does not decouple any longer ) jeopardises the unitarity of the model under study . in order to give a consistent interpretation and circumvent the unitarity problem , we propose the following physical scenario : the exchanged boson is relatively heavy @xmath174 , so that we are far under the threshold for the excitation of the unphysical longitudinal mode of @xmath175 as justified by the hierarchy of masses stated above .
we highlight two aspects concerning the potential specified by eqs . and :
* we are working in the low momentum - transfer limit and , according to the uncertainty principle , one has the interaction range approximated by @xmath176 . with this and the above hierarchy
, we have @xmath177 , which ensures that we are considering the potential for distances larger than the compton wavelengths of the @xmath114 particles that act as external currents .
this is consistent with the fact that the potential is a meaningful quantity for macroscopically large distances . * though our non - conserved pseudo - vector current couples to a massive vector boson
, the fact that we are working in a low - energy regime ( nr approximation , having in mind heavy @xmath178 and @xmath179 ) keeps us away from the danger of non - unitarity , which shows up in the high - energy domain @xmath180 , so we conclude that the contribution described by eq .
is physically meaningful .
the spin-1 @xmath163 potential has the form : @xmath181 + \nonumber \\ & + & \frac{(1+m_{a}r)}{r^{2}}\vec{l } \ , \cdot \left [ \frac{1 } { m_1 m_2 } \left ( \delta_{1 } \langle \vec{s}\rangle_{2 } + \delta_{2}\langle \vec{s}\rangle_{1 } \right ) + \ , \left ( \frac{\delta_1}{2m_2 ^ 2 } \langle \vec{s } \rangle_2 + \frac{\delta_2}{2m_1 ^ 2 } \langle \vec{s } \rangle_1 \right ) \right ] + \nonumber \\ & + & \frac{1}{m_{1}m_{2}r^2 } \left [ \langle \vec{s}\rangle_1 \cdot \langle \vec{s } \rangle_2 \ , \left ( 1 + m_{a}r + m_{a}^{2}r^{2 } \right ) - \left(\hat{r } \cdot \langle \vec{s } \rangle_{1}\right ) \left(\hat{r } \cdot \langle \vec{s } \rangle_{2}\right ) \ , \left ( 3 + 3m_{a}r + m_{a}^{2}r^{2 } \right ) \right ] + \nonumber \\ & - & \zeta \left [ \frac{\delta_1}{2m_2 ^ 2 r^2 } \left ( \delta_2 \left(1 + m_a r\right ) - \left ( 3 + 3m_{a}r + m_{a}^{2}r^{2 } \right ) \ , \left ( \hat{r } \cdot \vec{\epsilon}_2 \right ) \left ( \hat{r } \cdot \vec{\epsilon^\ast_2 } \right ) \right ) + \ , 1 \leftrightarrow 2 \ , \right ] \biggr\ } , \label{v_v_rep_vet } \end{aligned}\ ] ] where , as appears in eq . for @xmath182
, we have introduced the parameter @xmath127 .
again , its value in the vector field representation is @xmath183 , whereas @xmath184 in the case of the tensor field parametrization
. we shall be more specific about this point in the next section .
the electromagnetic potential for the spin-1 charge carriers correspond to the choice @xmath185 , with @xmath167 and @xmath168 denoting the electric charges of the particles . in this case
one clearly sees that the electromagnetic coulomb interaction dominates at leading order for both spin-1/2 and spin-1 currents and , at this level , their respective potentials are indistinguishable .
finally , the bosonic @xmath164 potential @xmath186 and here we recall that the @xmath48 current for @xmath114 is conserved , contrary to what happens with @xmath115 .
therefore , this potential does not present the problems of its fermionic counterpart . by comparing the results
, we notice very similar spin - dependence in both @xmath163 potentials , such as spin - orbit and spin - spin interactions
. however , the bosonic case has additional terms with monopole - monopole and the polarization - polarization interactions ( see the last term in eq . ) . from the fermionic @xmath164 potential , we observe an exclusive spin - spin interaction with a @xmath187dependence .
the bosonic case does not present these interactions and it reveals the same functional spin - dependence of the @xmath132 potential , eq .. to conclude this section , in possession of the potentials we have calculated for spin-1 charged currents / sources , we point out the possibility to apply our results to the study of the recently discovered heavy hyperhydrogen @xcite and hyperhelium @xcite . to study the details of the spectroscopy of the excited spin-1 states of @xmath188hypernuclei , a careful analysis of the spin dependence of the potentials is needed ( e.g. see ref .
@xcite ) and , in the particular case of the spin-1 states of the hypernuclei , such an analysis was actually lacking in the literature .
though we are not working out these applications in the present paper - it is out of our scope here - we endeavour to connect our results to the study of the excited spin-1/2 and spin-1 states of heavy hypernuclei .
we shall report on that elsewhere .
we have discussed above the spin-1 particle in terms of the familiar vector field representation , but this is not the only possibility @xcite@xcite . on - shell , a massive charged spin-1 particle may well be described by a complex 2-form gauge field as given below : @xmath189 with @xmath190 being the associated gauge - covariant 3-form field - strength .
the equations of motion are found to be @xmath191 following similar arguments as in the methodology , in the rest frame , we have @xmath192 where @xmath65 is the polarization 3-vector in the rest frame of the particle .
by suitably applying a lorentz boost we attain the final form for @xmath193 in the lab system : @xmath194\right\ } , \label{bij}\end{aligned}\ ] ] where the second term in eq . and the spatial component of eq .
can be shown to be analogous : it suffices to consider a specific direction and use the vector identity @xmath195 .
the normalization is given by @xmath196 .
to study the matter sources and currents for this particular representation , we state below the interaction lagrangian that contemplates all the couplings we are interested in : @xmath197\phi + \nonumber \\ & + & ig_{v}\left [ g^{\ast \ , \mu\nu\kappa}b_{\nu\kappa } - g^{\mu\nu\kappa}b^{\ast}_{\nu\kappa } + ( g-1 ) \partial^{\nu}\left(b_{\nu\alpha}b^{\ast \ , \mu\alpha } - b^{\ast}_{\nu\alpha } b^{\mu\alpha } \right ) \right]a_{\mu } + \nonumber \\ & + & ig_{pv } \left [ g^{\ast \mu \nu \kappa } \tilde{b}_{\nu \kappa } - g^ { \mu \nu \kappa } \tilde{b } _ { \nu \kappa}^{\ast } \right]a_{\mu } , \label{lag_g_bos_tens}\end{aligned}\ ] ] where the coupling constants have the following canonical mass dimensions : @xmath77 = [ g_{ps } ] = 1 $ ] and @xmath78 = [ g_{pv } ] = 0 $ ] .
the 3-form field - strength , @xmath198 , does not include the gauge - covariant derivative . as previously discussed in connection with the gauge current in the vector field representation ,
the vector current we present below is the one to order @xmath30 upon use of the subsidiary condition with the ordinary derivative , that is , the free subsidiary condition , which is compatible with our procedure to get the potential to order @xmath73 . following this prescription
, we obtained the same result as delgado - acosta _ et al _ @xcite in terms of the @xmath91 momentum variables , namely , @xmath199 where @xmath200\ ] ] is the spin generator of the tensor field representation and , from now on , the symbol @xmath201 shall denote the momentum - space sources / currents in the tensor field representation .
the nr sources / currents in this representation read : @xmath202 \right\ } \label{bs } \\
\mathcal{j}_{ps(s=1 ) } & = & -\frac{ig^1_{ps}}{2m_{1}^{2}}\vec{q}\cdot \langle \vec{s } \rangle_{1 } \label{ps2 } \\
\mathcal{j}^{0}_{v(s=1 ) } & = & e_1 \left\ { \delta_{1}\left ( 1 + \frac{\vec{p}^{\ , 2}}{2m_1 ^ 2 } + \frac{\vec{q}^{\ , 2}}{8m_1 ^ 2 } \right ) - \frac{e_1}{2m^2_1 } \left [ i(\vec{q } \times \vec{p } ) \cdot \langle \vec{s }
\rangle_1 + \delta_1 \vec{q}^{\ , 2 } - ( \vec{q } \cdot \vec{\epsilon}_1)\left ( \vec{q } \cdot \vec{\epsilon_1^\ast } \right ) \right ] \right\ } \label{bv } \\
\mathcal{j}^{i}_{v(s=1 ) } & = & e_1 \left [ \frac{\delta_{1}}{m_{1}}\vec{p}_{i } + \frac{i}{m_{1 } } \epsilon_{ijk}\vec{q}_{j}\langle \vec{s}_{k } \rangle_{1 } \right ] \label{bvi } \\
\mathcal{j}^{0}_{pv(s=1 ) } & = & -\frac{ig^1_{pv}}{2m_{1}}\vec{q}\cdot \langle \vec{s } \rangle_{1 } \\
\mathcal{j}^{i}_{pv(s=1 ) } & \sim & \mathcal{o}(v^{2 } ) . \label{bpv}\end{aligned}\ ] ] the sources / currents of the two different spin-1 representations are very similar in the spin sector .
basically , a global sign distinguishes one from the other in every case , but this sign does not interfere in the evaluations of the amplitude .
the only exception is the spatial component of the pseudo - vector current , where we notice some differences in the functional form . nevertheless , this detail is irrelevant , because its contribution to the @xmath164 or @xmath150 potentials leads to terms of order higher than two in the momenta , which we are currently ignoring in our calculations . for this reason ,
we refrain from giving the explicit ( lengthy ) form of @xmath203 and only indicate its order .
the remarkable differences appear in the polarization sector of the scalar source and time - component of the vector current .
we recall that , in the vector field representation , eqs . and
, we have obtained the following polarization - dependence : @xmath204 while in the tensor case , see eqs . and , we have @xmath205 according to these results , the two representations differ also due to opposite projections in the contribution coming from the polarization ; this could be associated with the particular representation of the spin-1 we are dealing with .
let us be more specific : in order to evaluate the nr limit of these particular cases , we had to take into account the weak " contributions ( those with momenta ) of the fields , such as the components @xmath206 , @xmath207 , see eqs . and , respectively , and the analogues present in @xmath61 and @xmath208 .
these different couplings between momenta and polarization , @xmath209 and @xmath210 , arise due to the lorentz boost to the lab frame and produce the observed longitudinal and transverse projections appearing in eqs . and above .
these results ( eqs . and ) raise a question concerning the vector or axial character of the spin-1 particle . in the @xmath211-case , we are assuming a vector - like particle . in the tensor representation ,
let us take the components as @xmath212 and @xmath213 . on - shell , in the rest frame of the particle , all the degrees of freedom ( d.f . )
are carried by @xmath214 .
going over to the lab - system , the tensor representation gives @xmath215 .
so , in a frame - independent way , the on - shell d.f . are actually described by @xmath216 .
the form of the vector current is insensitive to the vector or axial behavior of @xmath216 , as we may conclude by a close inspection of the @xmath217 current expressed in terms of @xmath218 and @xmath216 .
if @xmath216 is vector - like , then @xmath218 must be axial , and vice - versa .
therefore , we confirm that the minus sign difference highlighted above is always present as a consequence of the choice of representation ( vector or tensor ) , regardless of whether the particle is vector- or axial - like . at this point , it is instructive to discuss why we do not observe functional differences in all sources / currents , but only a global ( numerical or mass ) factor .
for example , by comparing the spatial component of the vector currents , eqs . and , we notice that the first contribution is of order @xmath219 .
one can check that the next correction should be @xmath220 , which we are neglecting in this work .
similar arguments apply to the other cases .
therefore , we could only achieve possible different profiles in these currents if we consider the relativistic corrections to higher orders . by using the same methodology as in section iii
it can be shown that the @xmath150 and @xmath164 potentials are identical to the results obtained in the vector field representation , eqs . and , while the @xmath133 and @xmath132 only differ by global mass factors , @xmath221 and @xmath222 , respectively , due to the normalization and the different canonical mass dimension of the coupling constant @xmath223 ( see eqs . and ) .
the other two potentials , @xmath113 and @xmath163 , have , respectively , the following profiles : @xmath224 \label{s_s_tensor}\ ] ] @xmath225.\ ] ] the difference between the two representations in both @xmath113 and @xmath163 cases appears as monopole - monopole terms and results in opposite behaviors in the polarization - polarization sector , since @xmath183 and @xmath226 in the vector and tensor representation , respectively . actually , these are consequences of what we have seen in the discussion above about the scalar * source * and time - component of the vector current . in general , we can not distinguish between the vector and tensor field representations for the @xmath114 by only considering the spin - dependent sector of the potentials .
we emphasize that the differences are suppressed because they only show up in order @xmath227 in the amplitude .
this is so because we are here bound to consider only dipole contributions .
if we extend our calculations to also include quadrupole effects , the differences in the @xmath163 potentials for the vector and tensor representation of the spin-1 might become evident , as we could expect from the studies reported in ref .
the consequences of the quadrupole moments of the two representations are interesting to be investigated and we intend to report on that in a forthcoming work @xcite .
in this paper , we focused our attention on the special properties of the spin - spin dependence of the source / current@xmath8source / current interaction potential in the non - relativistic limit for spin-1 and spin-1/2 charged matter particles .
our results indicate some universalities between bosons and fermions when exchanging scalar and vector particles .
essentially , we observe very similar contributions in the spin - sector of the @xmath113 , @xmath133 , @xmath132 and @xmath163 potentials .
the main differences show up in connection with @xmath48 currents , namely , the @xmath150 and @xmath164 potentials .
we highlight here a result presented in section iv : the non - trivial consequences of choosing a particular field representation for the spin-1 particle .
even if the vector and tensor representations are equivalent on - shell ( and the particles in the source / current are on - shell ) , the results for the @xmath113 and @xmath163 potentials do differ .
the justification was given in the aforementioned section and we claim that , by including higher - order terms in the relativistic corrections , quadrupole moments appear and the differences become explicit . furthermore , we do not present a discussion on the bosonic and fermionic ( pseudo-)tensor currents exchanging a 2-form gauge field .
however , according to preliminary results of a work in progress , it is worthwhile to comment that the case of bosonic sources / currents would yield interesting new interactions involving spin - polarization couplings , e.g. @xmath228 . to close this report , we point out a question that might be addressed to with the help of our results on the comparison between the classes of potentials describing the interaction between spin-1/2 and spin-1 matter sources / currents . for the fermionic case , the electromagnetic potential does not couple to the particle density , as it happens instead for spin-0 and spin-1 charged matter . if a charged spin-1 field , @xmath211 , is non - minimally coupled ( see eq . ) to the electromagnetic field , we have @xmath229 = \ ] ] @xmath230 where @xmath231 and @xmath232 .
the second term of the rhs contributing to @xmath233 stems from the non - minimal coupling from eq . and is mandatory to ensure @xmath46 for the charged spin-1 boson . from eq .
, two interesting properties of the electromagnetic interactions of spin-1 matter fields are made explicit : the london - type term that couples @xmath234 to the density of carriers and the new interaction between the photon and the charged boson through their polarization vectors .
these special aspects are expected to influence the inter - particle potential only if we take loop effects into account . however , the @xmath235terms in eq . introduce singularities associated to the point - like idealization of spin-1 charged matter particles . the 1962 paper by dirac @xcite and the recent works by fabbri @xcite and dain @xcite address the issue of extensibility in connection with charged particles .
in the present paper , we do not calculate classical field configurations generated by spin-1 currents ; instead , we build up interaction potentials by means of a semi - classical calculation , so we do not run into the complications yielded by the singularities mentioned above
. it would be nevertheless worthwhile to analyse the details of the connection between particle extension , mass , charge and spin .
the electrodynamics of bosonic carriers seems to suggest that the point - like idealization of charged particles is indeed very restrictive .
we hope to carry on with our studies in this direction and to make use of our results for the spin-1 potentials to get a deeper insight in the understanding of the electrodynamical properties of charged bosonic fields in the classical regime .
we would like to thank the national council for scientific and technological development of brazil ( cnpq ) , the co - ordination for qualification of higher education personnel ( capes ) , the german service for academic exchange ( daad ) and the fundao carlos chagas filho de amparo pesquisa do estado do rio de janeiro ( faperj ) for the invaluable financial support .
the authors also express their appreciation to the ( anonymous ) referees for the pertinent and constructive criticisms and suggestions on the original manuscript . | this paper sets out to establish a comparative study between classes of spin- and velocity - dependent potentials for spin-1/2 and spin-1 matter currents / sources in the non - relativistic regime .
both ( neutral massive ) scalar and vector particles are considered to mediate the interactions between ( pseudo-)scalar sources or ( pseudo-)vector currents .
though our discussion is more general , we contemplate specific cases in which our results may describe the electromagnetic interaction with a massive ( proca - type ) photon exchanged between two spin-1/2 or two spin-1 carriers .
we highlight the similarities and peculiarities of the potentials for the two different types of charged matter and also focus our attention on the comparison between the particular aspects of two different field representations for spin-1 matter particles .
we believe that our results may contribute to a further discussion of the relation between charge , spin and extensibility of elementary particles . |
the research of zo , gm and nc was partially supported by the shota rustaveli national science foundation grant ( n31/49 ) .
the work of sm was , in part , supported by usdoe contract no.de fg 03 - 96er-54366 .
school of physics , free university of tbilisi , 0183-tbilisi , georgia + zaza osmanov + institute for fusion studies , the university of texas at austin , austin , texas + swadesh mahajan + centre for theoretical astrophysics , itp , ilia state university , 0162-tbilisi , georgia + george machabeli , nino chkheidze zo conceived the project ; zo , sm , gm and nc developed the analytical model .
zo performed the numerical investigation of the problem , who also wrote the manuscript with the assistance of all other co - authors . | the newly born millisecond pulsars are investigated as possible energy sources for creating ultra - high energy electrons .
the transfer of energy from the star rotation to high energy electrons takes place through the landau damping of centrifugally driven ( via a two stream instability ) electrostatic langmuir waves .
generated in the bulk magnetosphere plasma , such waves grow to high amplitudes , and then damp , very effectively , on relativistic electrons driving them to even higher energies .
we show that the rate of transfer of energy is so efficient that no energy losses might affect the mechanism of particle acceleration ; the electrons might achieve energies of the order of @xmath0ev for parameters characteristic of a young star .
acceleration of particles .
* introduction * one of the most fascinating stories in astro - particle physics is the discovery of high energy cosmic ray ( cr ) electrons .
two important constituents of this story are : the h.e.s.s ( high energy stereoscopic system ) team s @xcite claim of the detection of tev cr electrons of local origin ( within @xmath1kpc ) , and the later confirmation of the energy range ( .007 - 1tev ) by the fermi - lat collaboration that analyzed the cr electron / positron data samples collected from @xmath2 august @xmath3 to @xmath2 august @xmath4 . @xcite
. how do particles acquire such enormous amounts of energy ?
in astro - particle physics , it is conventional to invoke a fermi - type process : the stochastic acceleration of particles interacting with strong shocks .
though quite effective , such a mechanism requires a pre acceleration of particles to already high energies @xcite .
accelerating light particles like electrons to high energies is even more problematic because the strong synchrotron radiation losses will limit the maximum attainable energy @xcite .
one can imagine electrons being boosted to pev or higher energies if strong energy losses could be , somehow , prevented during the accelerating process .
one such mechanism in which the rotation energy of a neutron star was utilized to strongly accelerate particles ( via langmuir waves generated in magnetospheres of pulsars ) , was recently developed and investigated .
it was shown the acceleration process proceeds with almost negligible energy losses .
@xcite .
the principle steps in the above mentioned process are : 1)the rotational slow down of a pulsar provides the energy to excite langmuir waves in the bulk electron - positron plasma in the star atmosphere .
all pulsars are characterized by a decreasing rate of spinning , @xmath5 , where @xmath6 is the period of rotation.the period is measured by @xmath7 , were @xmath8 is the moment of inertia of the neutron star , @xmath9 and @xmath10 g are the pulsar s mass and the solar mass respectively , @xmath11 cm is the pulsar s radius , @xmath12 is its angular velocity and @xmath13 .
\2 ) the excited langmuir waves , then , efficiently damp , preferentially on the much faster but local beam electrons to accelerate them to even larger energies we refer the reader to ( @xcite , and references therein ) for the relevant pulsar phenomenology , and details of this acceleration mechanism ; henceforth , the latter twould be termed langmuir - landau - centrifugal drive ( llcd ) .
llcd is based on the centrifugal acceleration that has been successfully applied to some astrophysical settings @xcite .
it is also believed that young pulsars might be sources of ultra high energy cosmic rays @xcite . for the paper to be self - sufficient
, we will give a synopsis of the theoretical underpinnings of llcd before applying it to the newly born millisecond pulsars ; the results are spectacular ; llcd can drive electron energies all the way to @xmath0ev . in this paper , we do not investigate the aftermath of the acceleration era- for instance , the observational patterns caused by the presence of ultra - high energy particles in the magnetospheres of millisecond pulsars
. one does expect that electrons with such enormous energies will leave their `` fingerprints '' on the emission patterns of the corresponding pulsars .
the scope of this work could be extended to examine the effects of the electromagnetic radiation that must come from the ultra relativistic electrons ; we intend to deal with these problems in near future .
another class of potentially interesting problems ( in the context of the present work ) are related to the recently discovered events of pev neutrinos by the ice cube experiment .
indeed , electrons and positrons with energies much higher than pev , might produce pev neutrinos via several channels : @xmath14 ; @xmath15 . *
results * the langmuir - landau - centrifugal drive , derived within the framework of a relatively simple but nontrivial theoretical model , is shown to work highly efficiently in the young millisecond pulsars .
llcd , through a two step process , converts the pulsar spin - down energy into the kinetic energy of electrons . in the first step ,
the rotation imparted centrifugal particle energy is converted , via a parametrically driven `` two stream instability '' , to electrostatic langmuir waves in the bulk electron - positron plasma residing in the pulsar magnetosphere .
landau damping of these centrifugally excited electrostatic waves on the high energy primary beam electrons ( beam ) transiting through the same region , constitutes the second step .
the linear growth rates of the rotation driven langmuir instability , calculated for the newly born pulsars , are faster than the typical rotation timescales of particles .
the langmuir instability , thus , is efficient and can rapidly convert the star slow down energy into the electric field energy landau damping of the unstable langmuir waves on primary beam electrons that converts the electric field energy into particle energy is also shown to be equally rapid .
the combination creates a very efficient `` machine '' that generates ultra high energy ( up to eev ) cosmic ray electrons .
we will show that the llcd mechanism , when applied to newly born fast spinning stars could , indeed , create electron energies that match the highest observed in the cr electrons ; we believe that this result has much significance for high energy astro - particle physics . * discussion
* it is generally believed that to attain the possible maximum possible energy , the gyroradius of the particle should be contained in the acceleration zone .
then , the combination of a strong rotation and a strong magnetic field ( of a pulsar ) leads to an enormous induced electric field on the light cylinder surface @xmath16statvolts / cm .
this field , might potentially accelerate electrons up to energies of the order of @xmath17ev @xcite , where @xmath18 , @xmath19 . what one needs
, then , is an efficient agency in the rotating magnetosphere that will convert the rotational energy into the acceleration of electrons .
we believe that the llcd does exactly that : the landau damping of centrifugally induced langmuir waves guarantees a high efficient channel for acceleration for these waves to effectively transfer their energy to particles , the waves phase velocity must be close to the particle speed , which , in this case , approaches the speed of light .
further , in the vicinity of @xmath20 , there should be more particles a little slower than the wave than particles which are a little faster ( in the opposite situation , the wave will feed off the particles ) . for the given problem it is always possible to situate @xmath20 in the desired part of the primary beam spectrum .
since the distribution function decreases with the lorentz factor , the number of electrons with @xmath21 exceeds that of the electrons with @xmath22 , where @xmath23 denotes the electron speed .
thus the optimum conditions for effective landau damping and , therefore , of net energy transfer from the star spin down to electrons , will pertain . for relativistic plasmas
it has been shown that the landau damping rate is given by @xcite @xmath24 where @xmath25 is the plasma density and @xmath26@xmath27 is the goldreich - julian number density . for the primary beam ,
@xmath28 and @xmath29 are , respectively , the lorentz factor and the plasma frequency .
landau damping is the second step of the acceleration mechanism llcd .
we now demonstrate the efficiency of llcd for young millisecond pulsars for which the typical period and slowdown rates are : @xmath30s and @xmath31ss@xmath32 .
it is worth noting that , at this stage such pulsars are only , theoretically , predicted @xcite ; this type of pulsars have not yet been observed .
the efficiency of the llcd must depend upon the proper coordination of the damping rates on the beam electrons and the growth rates of langmuir waves in the bulk plasma . in @xcite ,
the magnetospheric plasma was modeled as consisting of two streams with the lorentz factors @xmath33 ( @xmath34 ) .
it was shown that the growth rate of the instability is well approximated by the expression @xmath35 where @xmath36 are the plasma frequencies of the two species , @xmath37 are the corresponding number densities , @xmath38 is the bessel function of the first kind , @xmath39 , @xmath40 and @xmath41 and @xmath42 are , respectively , the positron s and electron s initial phases .
it stands to reason that if the instability growth rate was much greater than the landau damping rate , there will be little effective energy transfer from waves to the particles . in the diagonally opposite limit with the damping rates far in excess of the growth rates
, the waves will not grow much , again resulting in very little transfer from the star rotation to the waves . the most optimum scenario for an overall efficient energy pumping / transfer system ,
therefore , is realized when the instability growth and landau damping rates are large ( with respect to the kinematic rate , @xmath43 ) and comparable , @xmath44 . for the two - stream instability , the aforementioned condition can be satisfied for particular combinations of @xmath45 and @xmath46 ; for example , for @xmath47 , @xmath48 . the total energy gained by the beam particles has been estimated to be @xcite @xmath49 where @xmath50 , @xmath51 , @xmath52 @xcite . from eq .
( [ en ] ) it is clear that two streams with @xmath53 , @xmath48 , which in turn guarantee the condition , @xmath44 , lead to the acceleration of electrons up to energies of the order of @xmath0ev .
this estimate invokes equipartition of energy in the three relevant sp constituents : @xmath54 with @xmath55 , where @xmath56@xmath27 is the goldreich - julian number density @xcite in regions in the vicinity of the light cylinder .
because the highly relativistic particles will , inevitably , loose energy due to radiation , one must investigate how radiation will affect overall energy transfer .
could radiation , for example , put a stringent limit on the maximum energy acquired ? for highly relativistic electrons and photons with @xmath57 (
@xmath58 is photon energy ) , the inverse compton mechanism operates in the klein - nishina ( kn ) regime @xcite .
energy emitted per particle per second is , then , given by the approximate expression @xmath59 @xcite where @xmath60 is the classical electron radius .
the corresponding time scale of the process is @xmath61 . using the approximate photon number density , @xmath62 , and taking into account the typical values of luminosity in the high energy level , @xmath63gev , satisfy @xmath64erg / s , it is straightforward to show that for energies of our interest @xmath65pev , the aforementioned timescale exceeds the instability time scale by many orders of magnitude .
thus compton cooling is very slow compared to the wave energy transfer time .
the kn time scale goes up with the particle energy implying that at higher energies , the inverse compton mechanism in the kn channel is too slow to impose any constraints on the maximum attainable electron energy .
the next possible loss mechanism for relativistic particles , moving in magnetic field , is the synchrotron emission with the corresponding estimated power @xmath66 @xcite , where @xmath67 is the cyclotron frequency . according to the standard model of pulsar magnetospheres
, there is a significant region over the star s surface , where the electric field is nonzero ( vacuum gap ) .
it is this electric field that , first , accelerates electrons to relativistic energies @xcite .
one can readily show that the electrons leaving the gap with a @xmath68 will undergo efficient synchrotron cooling at a short timescale , @xmath69s .
thus , immediately after leaving the gap , the particles radiate their transverse momentum , and very soon transit to the ground landau state .
after that , zipping only along the field lines , the electrons will reach the light cylinder zone in due course of time .
it is precisely the region , where the langmuir waves , always propagating along the field lines , are excited .
therefore the wave interaction with particles will not cause pitch angle scattering , efficiently suppressing the synchrotron mechanism .
quasi linear diffusion , another possible source for imparting a pitch angle , @xcite , also does not operate because the required condition for diffusion @xmath70 , is violated for extremely energetic plasmas ( for plasmas with energy density exceeding that of magnetic field ) .
the synchrotron mechanism is not expected to impose any constraints on the maximum attainable energies .
how about the curvature radiation emitted by particles moving along the curved magnetic field lines ?
to explain force free regime of particles leaving a pulsar s magnetosphere , we have tried to reconstruct the structure of magnetic field close to the light cylinder surface @xcite .
we showed that shown that the curvature driven current imparts a toroidal component to the magnetic field . as a result ,
the field lines lag behind the rotation , gradually erasing the instability .
it has been shown that the corresponding timescale is approximately given by @xmath71^{-1},\ ] ] where @xmath72 is the curvature drift velocity , @xmath73 is the curvature radius of a magnetic field line and @xmath74 and @xmath75 are the components of the wave vector of the induced curvature drift mode respectively along the drift and perpendicular to it .
one can see that for a wide range of parameters corresponding to the drift mode with small inclination angles @xmath76 ( the angle of the wave vector with respect to the drift direction ) , the instability timescale is less than the kinematic timescale @xmath6 .
in particular , for @xmath77 and @xmath78 , @xmath79 . therefore , the drift instability is so efficient that the magnetic field lines very rapidly reconstruct resulting in a configuration , that enables particles to follow straight line trajectories . the curvature radiation is minimized and does not quite interfere with our energy transfer mechanism . * methods
* the llcd was developed using the standard model of the pulsar magnetosphere , where the particle distribution can be , schematically , represented as shown in fig .
[ fig1 ] . the narrower region on the figure describes the primary beam electrons with higher lorentz factors ( @xmath28 ) . the relatively wider sector with lorentz factors in the interval @xmath80 $ ] ( @xmath81 ) characterizes the secondary particles produced by means of pair creation .
such a shape of the distribution function makes the development of two stream instabilities feasible in the magnetospheres of pulsars @xcite .
this feature of the electron distributions , along with interesting orbit effects , was exploited in @xcite to demonstrate the the existence of a strong two - stream like instability that is parametrically driven to excite high levels of langmuir waves . in the framework of @xmath82 formalism @xcite ,
the basic system of the euler equation and continuity equations ( for each species ) and the poisson equation , is adequate in describing the centrifugally - excited ( rotationally driven ) langmuir waves .
the fourier transformed linear system consists of @xmath83 @xmath84 @xmath85 where @xmath86 is the species index ( electrons and positrons ) , @xmath87 and @xmath88 are , respectively , the first order dimensionless momentum ( @xmath89 ) and the zeroth order velocity , @xmath90 is the electron mass , @xmath91 is the speed of light , @xmath92 is the charge of the corresponding particle , @xmath93 is the radial coordinate of the corresponding specie , @xmath94 and @xmath95 are respectively the perturbed and nonperturbed fourier components of the density , @xmath96 is the wave number and @xmath97 is the electrostatic field .
the first term in the righthand side of the euler equation comes from the so - called centrifugal force .
the centrifugal force , controlled by the following `` orbits '' of initially relativistic particles @xmath98 , @xmath99 @xcite , where @xmath100 denotes the initial phase of a given species , is time dependent .
the resulting langmuir mode equation is periodic in time ( has the generalized mathieu / hill form ) and exhibits parametric instability .
the time dependent centrifugal force that parametrically drives the electrostatic waves is different for the two species - so are their lorentz factors . since @xmath101 ,
the order of the bessel function is rather high , non - zero growth rates will pertain only if the argument @xmath102 is comparable to @xmath101 @xcite . for the instability to be called `` efficient '' , the inverse growth time @xmath103 should not exceed the kinematic timescale (
also called the escape timescale ) , @xmath104 . in fig .
[ fig2 ] , we plot the ratio @xmath105 versus @xmath46 for different values of @xmath45 : @xmath106 ( dashed - dotted line ) ; @xmath107 ( dashed line ) and @xmath108 ( solid line ) .
the results are derived for typical young millisecond pulsars with @xmath109s and @xmath110 , where @xmath111 g is the magnetic field close to the pulsar s surface and @xmath112 is the light cylinder radius . @xmath113ss@xmath32 . the condition @xmath114 is satisfied for a broad range of lorentz factors ; the instability is , indeed , efficient .
thus we have reviewed , for the young millisecond pulsars , the first step of llcd : the existence of an efficient centrifugally excited parametric instability that pumps the spin down energy of the rotator into electrostatic langmuir waves .
now we review how , through landau damping , this energy will be conveyed to the fastest of the particles .
the landau damping in pulsar magnetospheres has been investigated in @xcite , but the authors focus on alfven waves , but not on the electrostatic waves .
999 aharonian , f. , et al . , energy spectrum of cosmic - ray electrons at tev energies .
_ phys .
rev .
l. _ , * 101 * , 261104 , ( 2008 ) ackermann , m. , et al .
fermi lat observations of cosmic - ray electrons from 7 gev to 1 tev .
_ phys .
rev .
d. _ , * 82 * , 092004 , ( 2010 ) bell , a. r. , the acceleration of cosmic rays in shock fronts .
_ mnras _ , * 182 * , 147 - 156 , ( 1978 ) blandford , r. d. , netzer , h. & woltjer , l. _ active galactic nuclei _ , springer - verlag ( 1990 ) mahajan , s. , machabeli , g. , osmanov , z. & chkheidze , n. , ultra high energy electrons powered by pulsar rotation .
_ nat .
sci .
rep . _ * 3 * , 1262 , ( 2013 ) bogovalov , s. , magnetocentrifugal acceleration of plasma in a nonaxisymmetric magnetosphere .
_ a&a _ , * 367 * , 159 - 169 , ( 2001 ) osmanov , z. , rogava , a.s . & bodo , g. , on the efficiency of particle acceleration by rotating magnetospheres in agn .
_ a&a _ , * 470 * , 395 - 400 , ( 2007 ) fang , k. , kotera , k. & olinto , a. v. , newly born pulsars as sources of ultrahigh energy cosmic rays .
_ apj _ , * 750 * , 1 - 16 , ( 2012 ) volokitin , a.s . ,
krasnoselskikh , v.v . & machabeli , g.z . , waves in the relativistic electron - positron plasma of a pulsar .
_ svjpp _ , * 11 * , 310 - 314 , ( 1985 ) carroll , b.w .
& ostlie , d.a .
_ an introduction to modern astrophysics _ , pearson international edition ( 2006 ) rogava , a. , dalakishvili , g. & osmanov , z. , centrifugally driven relativistic dynamics on curved trajectories .
_ gregr _ , * 35 * , 1133 - 1152 , ( 2003 ) goldreich , p. & julian , w. h. , pulsar electrodynamics .
_ apj _ , * 157 * , 869 - 880 , ( 1969 ) rybicki , g.b .
& lightman , a. p. , _ radiative processes in astrophysics_. wiley , new york , ( 1979 ) blumenthal , g. r. & gould , r. j. , bremsstrahlung , synchrotron radiation , and compton scattering of high - energy electrons traversing dilute gases . _
rev .
mod .
phys .
_ , * 42 * , 237 - 271 , ( 1970 ) ruderman , m. a. & sutherland , p. g. , theory of pulsars - polar caps , sparks , and coherent microwave radiation .
_ apj _ , * 196 * , 51 - 72 , ( 1975 ) lominadze j.g .
, machabeli g.z . &
mikhailovsky a.b . ,
1979 , relativistic electron - positron plasma quasi - linear relaxation at the presence of magnetic bremsstrahlung .
_ j. phys .
colloq .
_ , * 40 * , no .
c-7 , 713 , ( 1979 ) osmanov , z. , shapakidze , d. & machabeli g. , dynamical feedback of the curvature drift instability on its saturation process .
_ a&a _ , * 503 * , 19 - 24 , ( 2009 ) gedalin , m. , gruman , e. & melrose d. b. , new mechanism of pulsar radio emission .
_ phys .
rev .
l. _ , * 88 * , 121101 , ( 2002 ) thorne , k. , price , r. & macdonald , d.a .
_ black holes : the membrane paradigm _ , yale university press , new haven ( 1986 ) abramowitz , m. & stegun , i. a. , _ handbook of mathematical functions _ , edited by abramowitz , m. & stegun , i. a. , natl
bur .
stand .
appl .
math .
ser .
no .
55 ( u.s .
gpo , washington , d.c . , 1965 ) arons , j. & scharlemann , e. t. , pair formation above pulsar polar caps - structure of the low altitude acceleration zone .
_ apj _ , * 231 * , 854 - 879 , ( 1979 ) |
in biological systems , the survival of a species depends on the frequencies of its kin and its foes in the environment @xcite . in some cases
, the chance of survival of a certain species _ improves _ as the frequency of its kind increases , since this might enhance the chance for reproduction or other benefits from group interaction .
this is denoted as _ positive _ frequency dependence . in other cases
a _ negative _ frequency dependence , that is the increase of the survival chance with _ decreasing _ frequency , is observed .
this is the case , when individuals compete for rare ressources .
moreover , negative frequency dependence is known to be important for maintaining the genetic diversity in natural populations @xcite .
frequency dependent dynamics are not only found in biological systems , but also in social and economic systems @xcite . in democracies , a simple example is a public vote , where the winning chances of a party increase with the number of supporters @xcite . in economics ,
e.g. the acceptance of a new products may increase with the number of its users @xcite . in stock markets ,
on the other hand , positive and negative frequency dependencies may interfere .
for instance , the desire to buy a certain stock may increase with the orders observed from others , a phenomenon known as the _ herding effect _ , but it also may decrease , because traders fear speculative bubbles . in general
, many biological and socio - economic processes are governed by the frequency dependent adoption of a certain behavior or strategy , or simply by frequency dependent reproduction . in order to model such dynamics more rigorously ( but less concrete ) , different versions of _ voter models _ have been investigated .
the voter model denotes a simple binary system comprised of @xmath1 _ voters _ , each of which can be in one of two states ( where _ state _ could stand for opinion , attitude , or occupation etc . ) , @xmath2 . here
, the transition rate @xmath3 from state @xmath4 to state @xmath5 is assumed to be proportional to the frequency @xmath6 . in this paper , we extend this approach by assuming a _ nonlinear _ voter model , where the frequency dependence of the transition rate , @xmath7 , includes an additional nonlinearity expressed in terms of the ( frequency dependent ) prefactor @xmath8 .
linear voter models have been discussed for a long time in mathematics @xcite .
recently , they gained more attention in statistical physics @xcite because of some remarkable features in their dynamics described in sect .
[ sec : vm ] .
but voter models also found the interest of population biologists @xcite dependent on how the frequency @xmath6 is estimated , one can distinguish global from local voter models . in the latter case
the transition is governed only by the local frequency of a certain state in a given neighborhood .
in contrast to global ( or mean - field ) models , this leads to local effects in the dynamics , which are of particular interest in the current paper .
if space is represented by a two - dimenional lattice and each site is occupied by just one individual , then each species occupies an amount of space proportional to its presence in the total population .
local effects such as the occupation of a neighborhood by a particular species or the adoption of a given opinion in a certain surrounding , can then be observed graphically in terms of domain formation .
this way , the invasion of species ( or opinions ) in the environment displays obvious analogies to spatial pattern formation in physical systems .
physicists have developed different spatial models for such processes .
one recent example is the so - called `` sznajd model '' @xcite which is a simple cellular automata ( ca ) approach to consensus formation ( i.e. complete invasion ) among two opposite opinions ( described by spin up or down ) . in @xcite , we have shown that the sznajd model can be completely reformulated in terms of a linear voter model , where the transition rates towards a given opinion are directly proportional to the frequency of the respective opinion of the _ second - nearest _ neighbors and independent of the nearest neighbors . other spatial models are proposed for game - theoretical interactions among nearest neighbors @xcite . here
, the dynamics are driven by local payoff differences of adjacent players , which basically determine the nonlinearity @xmath9 .
dependent on these payoff differences , we could derive a phase diagram with five regimes , each characterized by a distinct spatio - temporal dynamic @xcite .
the corresponding spatial patterns range from complete invasion to coexistence with large domains , coexistence with small clusters , and spatial chaos . in this paper
, we are interested in the local effects of frequency dependent dynamics in a homogeneous network , where each site has @xmath0 nearest neighbors . in this case
, the nonlinearity @xmath9 can be simply expressed by two constants , @xmath10 , @xmath11 .
this is a special form of a nonlinear voter model , which for @xmath12 also includes majority voting and for @xmath13 minority voting .
we investigate the dynamics of this model both analytically and by means of computer simulations on a two - dimensional stochastic ca ( which is a special form of a homogeneous network with @xmath0 ) .
the latter one was already studied in @xcite , in particular there was a phase diagram obtained via numerical simulations . in our paper , we go beyond that approach by deriving the phase diagram from an analytical approximation , which is then compared with our own simulations . in sects .
[ sec : formal ] , [ sec : micro ] we introduce the microscopic model of frequency dependent invasion and demonstrate in sects .
[ sec : a1a2 ] , [ 2.4 ] the role of @xmath10 , @xmath11 by means of characteristic pattern formation . based on the microscopic description , in sect .
[ 3.1 ] we derive the dynamics for the global frequency @xmath14 , which is a macroscopic key variable .
an analytical investigation of these dynamics is made possible by pair approximation , sect .
[ 3.2 ] , which results in a closed - form description for @xmath14 and the spatial correlations @xmath15 . in sect .
[ 4.1 ] , we verify the performance of our analytical approximations by comparing them with averaged ca computer simulations .
the outcome of the comparison allows us to derive in sect .
[ 4.3 ] a phase diagram in the @xmath16 parameter space , to distinguish between two possible dynamic scenarious : ( i ) complete invasion of one of the species , with formation of domains at intermediate time scales , and ( ii ) random spatial coexistence of two species .
a third dynamic regime , the nonstationary coexistence of the two species on long time scales together with the formation of spatial domains , can be found in a small , but extended region that separates the two dynamic regimes mentioned above .
we further discuss in sect . [ 5 ] that the usual distinctions for the dynamics , such as positive or negative frequency dependence , do not necessarily coincide with the different dynamic regimes . instead , for positive frequency dependence ,
all of the three different dynamic regimes ( and the related spatio - temporal patterns ) are observed . in the appendix , calculation details for the pair approximation
are given .
we consider a model of two species labeled by the index @xmath17 .
the total number of individuals is constant , so the global frequency @xmath18 ( or the share of each species in the total population ) is defined as : @xmath19 in the following , the variable @xmath20 shall refer to the global frequency of species 1 .
the individuals of the two species are identified by the index @xmath21 and can be seen as nodes of a network .
a discrete value @xmath22 indicates whether the node is occupied by species 0 or 1 .
the network topology ( specified later ) then defines the nearest neighbors @xmath23 of node @xmath24 . in this paper , we assume homogeneous networks where all nodes have the same number @xmath25 of nearest neighbors . for further use , we define the local occupation @xmath26 of the nearest neighborhood ( without node @xmath24 ) as : @xmath27 a specific realization of this distribution shall be denoted as @xmath28 , while the function @xmath29 assigns @xmath28 to a particular neighborhood @xmath26 : @xmath30 for later use , it is convenient to define these distributions also for the nearest neighborhood _ including _
node @xmath24 : @xmath31 for @xmath0 , @xmath32 denotes a binary string , e.g. @xmath33 , where the first value @xmath34 refers to the center node and the other values @xmath35 indicate the particular values of the nearest neighbors .
the assignment of these values to a particular neighborhood @xmath36 of node @xmath24 is then described by @xmath37 . in the voter model described in the following section , the dynamics of @xmath38
is governed by the _ occupation distribution _ of the _ local neighborhood_. that surrounds each node @xmath24 .
using a stochastic approach , the probability @xmath39 to find node @xmath24 in state @xmath38 therefore depends in general on the local occupation distribution @xmath26 of the neigborhood ( eqn .
( [ occupat ] ) , in the following manner : @xmath40 hence , @xmath39 is defined as the marginal distribution of @xmath41 , where @xmath42 in eqn .
( [ eq : marg_dist ] ) indicates the summation over all possible realizations of the local occupation distribution @xmath26 , namely @xmath43 different possibilities . for the time
dependent change of @xmath39 we assume the following master equation : @xmath44\end{aligned}\ ] ] where @xmath45 denotes the transition rate for state @xmath46 of node @xmath24 into state @xmath38 in the next time step under the condition that the local occupation distribution is given by @xmath26 . the transition rate for the reverse process is @xmath47 .
again , the summation is over all possible realizations of @xmath26 , denoted by @xmath42 .
it remains to specify the transition rates , which is done in the following section .
our dynamic assumptions for the change of an individual state @xmath38 are taken from the so - called voter model ( see also sect . [ 1 ] ) , abbreviated as vm in the following .
the dynamics is given by the following update rule : a voter , i.e. a node @xmath48 of the network , is selected at random and adopts the state of a randomly chosen nearest neighbor @xmath49 .
after @xmath1 such update events , time is increased by 1 .
the probability to choose a voter with a given state @xmath34 from the neighborhood @xmath23 of voter @xmath24 is directly proportional to the relative number ( or frequency ) of voters with that particular state in that neigborhood .
let us define the _ local frequencies _ in the neighborhood as : @xmath50 where @xmath51 is the kronecker delta , which is 1 only for @xmath52 and zero otherwise .
then the transition rate of voter @xmath24 to change its state @xmath38 does not explicitly depend on the local distribution @xmath26 , but only on the _ occupation frequency _ @xmath53 , i.e. on the number of nodes occupied by either 0 or 1 in the neighborhood of size @xmath25 .
hence , the vm describes a frequency dependent dynamics : the larger the frequency of a given state in the neighborhood , the larger the probability of a voter to switch to that particular state _ if _ it is not already in that state .
i.e. the transition rate @xmath54 , to _ change _ state @xmath5 increases only with the local frequency of _ opposite _ states , @xmath55 , in the neighborhood : @xmath56 the prefactor @xmath57 determines the time scale of the transitions and is set to @xmath58 .
( [ eq : linvm ] , describes the dynamics of the _ linear _ vm because , according to the above update rule , the rate to _ change _ the state is directly proportional to the frequency . the linear ( or standard ) vm has two remarkable features .
first , it is known that , starting from a random distribution of states , the system always reaches a completely ordered state , which is often referred to as _ consensus _ in a social context , or complete _ invasion _ in a population biology context . as there are individuals with two different states , the complete ordered state can be either all 0 or all 1 . which of these two possible attractors of the dynamics is eventually reached , depends ( in addition to stochastic fluctuations ) on the initial global frequency , i.e. @xmath59 .
it has been shown that , for an ensemble average , the frequency of the outcome of a particular consensus state @xmath60 is equal to the initial frequency @xmath59 of state @xmath60 .
this second remarkable feature is often denoted as conservation of magnetization , where the magnetization is defined as @xmath61 .
hence , consensus means @xmath62 .
thus we have the interesting situation that , for a single realization , the dynamics of the linear vm is a fluctuation driven process that , for finite system sizes , always reaches consensus , whereas on average the outcome of the consensus state is distributed as @xmath63 .
the ( only ) interesting question for the linear vm is then how long it may take the system to reach the consensus state , dependent on the system size @xmath1 and the network topology .
the time to to reach consensus , @xmath64 , is obtained through an average over many realizations . as the investigation of @xmath64 is not the focus of our paper ( see @xcite ) ,
we just mention some known results for the linear vm : one finds for one - dimensional regular lattices ( @xmath65 ) @xmath66 and for two - dimensional regular lattices ( @xmath67 ) @xmath68 . for @xmath69
the system does not always reach an ordered state in the thermodynamic limit .
in finite systems , however , one finds @xmath70 .
while the linear vm has some nice theoretical properties , it also has several conceptual disadvantages when applying the model to a social or population biological context .
first of all , the `` voters '' do not vote in this model , they are subject to a random ( but frequency based ) assignment of an `` opinion '' , without any choice . secondly , the state of the voter under consideration does not play any role in the dynamics .
this can be interpreted in a social context as a ( blind ) herding dynamics , where the individuals just adopt the opinion of the majority . in a population model of two competing species , it means that individuals from a minority species may be replaced by those from a majority species without any resistance . in order to give voter @xmath24 at least some weight compared to the influence of its neighbors @xmath23 , one can simply count its state @xmath38 into the local frequency @xmath53 , i.e. instead of eqs .
( [ occupat ] ) , ( [ sigm01 ] ) we may use eqn . ( [ occupat0 ] ) . using for voter @xmath24 the notation @xmath71 ( i.e @xmath72 )
, we can still use eqn .
( [ eq : linvm ] ) for the transition rates , with the noticable difference that the local frequency @xmath53 of eqn .
( [ sum ] ) is now calculated from a summation that starts with @xmath73 .
the explicit consideration of @xmath38 thus has the effect of adding some _ inertia _ to the dynamics .
in fact , extending the summation to @xmath73 multiplies the transition rate , eqn .
( [ eq : linvm ] ) , by a factor @xmath74 , where @xmath25 is the number of nearest neighbors .
i.e. , for @xmath0 a local configuration @xmath75 would lead to a transition rate @xmath76 _ without _ the additional inertia , but @xmath77 by counting in the state of voter @xmath24 .
i.e. , taking into account the state of voter @xmath24 considerably reduces the transition rate towards the opposite state .
we find it useful for conceptual reasons to include some resistance into the model and therefore will use from now on the description which takes the current state of voter @xmath24 into account .
this also has the nice advantage that for the case @xmath0 , which describes e.g. square lattices , we avoid stalemate situations , @xmath78 .
however , we note that the addition of the constant resistance does not change the dynamics of the model , as it only adjusts the _ time scale _ towards a new factor @xmath79 .
so , keeping @xmath25 constant and equal for all voters , we can rescale @xmath80 .
we note that there are of course other ways to give some weight to the opinion of voter @xmath24 . in @xcite , we have discussed a modified vm , where voters additionally have an inertia @xmath81 $ ] which leads to a decrease of the transition rate to change their state : @xmath82 here @xmath83 is given by the linear vm , eqn .
( [ eq : linvm ] ) .
the individual inertia @xmath84 is evolving over time by assuming that it increases with the persistence time @xmath85 the voter has been keeping its current state .
while this intertia may slow down the microscopic dynamics of the vm and thus may increase the time to reach consensus , @xmath64 , we found the counterintuitive result that under certain circumstances a decelerated microdynamics may even accelerate the macrodynamics of the vm , thus decreasing @xmath64 compared to the linear vm .
the addition of a nonlinear inertia to the vm , eqn .
( [ eq : reluc ] ) , is a special case for turning the linear vm into a nonlinear one ( wheras the fixed resistence would not change the linear vm ) . in general ,
nonlinear vm can be expressed as @xmath86 where @xmath9 is a nonlinear , frequency dependent function describing how voter @xmath24 reacts on the occurence of opposite `` opinions '' in its immediate neighborhood .
[ fig : nonlin ] shows some possible examples which have their specific meaning in a social context .
whereas any function @xmath87 describes the linear vm , i.e. a majority voting or herding effect , a decreasing @xmath9 means minority voting , i.e. the voter tends to adopt the opinion of the minority .
nonmonotonous @xmath9 can account for voting against the trend , i.e. the voter adopts an opinion as long as this is not already the ruling opinion a phenomenon which is important e.g. in modeling the adoption of fashion .
an interpretation of these functions in a population biology context will be given in sect .
[ sec : a1a2 ] in conclusion , introducing the nonlinear response function @xmath9 will allow us to change the global dynamics of the linear vm . instead of reaching always consensus , i.e. the exclusive domination of one `` opinion '' or species
, we may be able to observe some more interesting macroscopic dynamics , for example the coexistence of both states .
it is one of the aims of this paper to find out , under which specifications of @xmath9 we may in fact obtain a dynamic transition that leads to a a structured , but not fully ordered state instead of a completely ordered state . ) .
note the piecewise linear functions , as the number of neighbors @xmath25 and thus the frequencies @xmath88 have mostly discrete values.,width=264 ]
in order to give a complete picture of the dynamics of the nonlinear vm , we have to derive the stochastic dynamics for the _ whole _ system of @xmath1 nodes , whereas eqn .
( [ master ] ) gives us `` only '' the _ local _ dynamics in the vicinity of a particular voter @xmath24 .
for @xmath1 voters , the distribution of states is given by @xmath89 note that the state space @xmath90 of all possible configurations is of the order @xmath91 . in a stochastic model
, we consider the probability @xmath92 of finding a particular configuration at time @xmath93 .
if @xmath93 is measured in discrete time steps ( generations ) and the network is synchronously updated , the time - dependent change of @xmath92 is described as follows : @xmath94 where @xmath95 denotes all possible realizations of @xmath96 and @xmath97 denote the conditional probabilities to go from state @xmath95 at time @xmath93 to @xmath96 at time @xmath98
( [ eq : t+1 ] ) is based on the markov assumption that the dynamics at time @xmath98 may depend only on states at time @xmath93 . with the assumption of small time steps @xmath99 and the definition of the transition rates @xmath100 eqn .
( [ eq : t+1 ] ) can be transferred into a time - continuous master equation as follows : @xmath101\ ] ] in eqn .
( [ master0 ] ) , the transition rates depend on the _ whole _ distribution @xmath96 .
however , in the frequency dependent dynamics introduced in sect .
[ 2.1 ] , only the occupation distribution of the _ local _ neighborhood of node @xmath24 needs to be taken into account .
therefore , it is appropriate to think about some reduced description in terms of lower order distributions , such as the local occupation @xmath26 , eqn .
( [ occupat ] ) . in principle
, there are two different ways to solve this task . the first one , the _ top - down _ approach starts from the _ global _ distribution @xmath96 in the whole state space and then uses different approaches to factorize @xmath92 .
however , a markov analysis @xcite can only be carried out exactly for small @xmath1 , because of the exponential @xmath1-dependence of the state space .
thus , for larger @xmath1 suitable approximations , partly derived from theoretical concepts in computer science need to be taken into account . in this paper ,
we follow a second way which is a _ bottom - up _ approach based on the _ local _ description already given in sect . [ 2.1 ] .
i.e. starting from node @xmath24 and its local neighborhood , we want to derive the dynamics for some appropriate _ macroscopic _ variables describing the nonlinear vm . instead of one equation for @xmath92 in the top - down approach , in the bottom - up
approach we now have a set of @xmath1 stochastic equations for @xmath39 , eqn .
( [ master0 ] ) , which are locally coupled because of overlapping neighborhoods , @xmath26 . in order to solve the dynamics
, we need to discuss suitable approximations for these local correlations .
as we are interested in the macroscopic dynamics , these approximations will be done at the macroscopic level . in order to do so
, we first derive a macroscopic equation from the stochastic eqn .
( [ master ] ) , which is carried out in the following section .
the key variable of the macroscopic dynamics is the global frequency @xmath102 , defined in eqn .
( [ nconst ] ) . in order to compare the averaged computer simulations with results from analytical approximations later in sect . [ 4 ] , we first derive an equation for the expectation value @xmath103 .
we do this without an explicit determination of the transition rates and wish to emphasize that the formal approach presented in sect .
[ 2.2 ] remains valid not just for the voter model , but also for other dynamic processes which depend on neighbor interactions ( not only nearest neighbors ) in various network topologies .
for the derivation of the expectation value we start from the stochastic description given in sect . [
sec : micro ] , where @xmath104 denoted the probability to find a particular distribution @xmath105 , eqn .
( [ vector ] ) , at time @xmath93 and @xmath106 denoted all possible realizations of @xmath105 eqn .
( [ master0 ] ) . on one hand : @xmath107 and on the other hand : @xmath108 by differentiating eqn .
( [ av2 ] ) with respect to time and inserting the master eqn .
( [ master ] ) , we find the following macroscopic dynamics for the network : @xmath109\end{aligned}\ ] ] for the further treatment of eqn .
( [ eq : master_macro1 ] ) , we consider a specific distribution of states on @xmath110 nodes defined by @xmath32 .
this distribution is assigned to a particular neighborhood of node @xmath24 by @xmath37 ( eqn .
( [ occupat0 ] ) ) .
since we are interested in how many times a special realization of a specific distribution @xmath111 is present in the population , we define an indicator function @xmath112 that is @xmath60 if the neighborhood of node @xmath24 has the distribution @xmath111 , and @xmath113 otherwise . therefore , we write the frequency of the @xmath114-tuplet @xmath32 in the population as : @xmath115 the expectation value is @xmath116 inserting eqn .
( [ eq : x_ind ] ) into eqn .
( [ eq : av_x ] ) , we verify that @xmath117 because of the definition of the marginal distribution . using the identity @xmath118 , we may rewrite eqn .
( [ eq : master_macro1 ] ) by means of eqn .
( [ eq : av_x1 ] ) to derive the macroscopic dynamics in the final form : @xmath119 \end{aligned}\ ] ] @xmath120 denotes the @xmath43 possible configurations of a specific occupation distribution @xmath28 , eqn .
( [ sigm01 ] ) ) . in the following , we use @xmath121 .
then , the dynamic for @xmath122 reads : @xmath123 \end{aligned}\ ] ] the solution of eqn .
( [ x - fin ] ) would require the computation of the averaged global frequencies @xmath124 and @xmath125 for all possible occupation patterns @xmath28 , which would be a tremendous effort .
therefore , in the next section we will introduce two analytical approximations to solve this problem . in sect . [
4.1 ] we will further show by means of computer simulations that these approximations are able to describe the averaged dynamics of the nonlinear vm . as a first approximation of eqn .
( [ x - fin ] ) , we investigate the mean - field limit . here the state of each node does not depend on the occupation distribution of its neighbors , but on @xmath25 randomly chosen nodes . in this case
the occupation distribution factorizes : @xmath126 for the macroscopic dynamics , eqn .
( [ x - fin ] ) , we find : @xmath127 \end{aligned}\ ] ] for the calculation of the @xmath128 we have to look at each possible occupation pattern @xmath28 for a neighborhood @xmath25 .
this will be done in detail in sect .
[ sec : a1a2 ] .
before , we discuss another analytical approximation which solves the macroscopic eqn .
( [ x - fin ] ) with respect to _
this is the so - called _ pair approximation _
, where one is not interested in the occupation distribution of a whole neighborhood @xmath32 , eqn .
( [ occupat0 ] ) ) but only in _ pairs _ of nearest neigbor nodes , @xmath129 with @xmath130 .
that means the local neighborhood of nearest neighbors is decomposed into pairs , i.e. blocks of size 2 that are called _ doublets_. similar to eqn .
( [ eq : x_ind ] ) , the global frequency of doublets is defined as : @xmath131 the expected value of the doublet frequency is then given by @xmath132 in the same way as in eqn .
( [ eq : av_x ] ) .
we now define the correlation term as : @xmath133 neglecting higher order correlations .
thus @xmath134 can be seen as an approximation of the conditional probability that a randomly chosen nearest neighbor of a node in state @xmath135 is in state @xmath34 . using the above definitions
, we have the following relations : @xmath136 for the case of two species @xmath137 , @xmath138 and @xmath139 are the _
inter - species _ correlations , while @xmath140 and @xmath141 denote the _ intra - species _ correlations . using @xmath142 ,
these correlations can be expressed in terms of only @xmath138 and @xmath122 as follows : @xmath143 now , the objective is to express the global frequency of a specific occupation pattern @xmath144 , eqn .
( [ eq : av_x ] ) , in terms of the correlation terms @xmath134 . in pair approximation
, it is assumed that the states @xmath145 are correlated only through the state @xmath146 and uncorrelated otherwise
. then the global frequency terms in eqn .
( [ eq : master_macrof ] ) can be approximated as follows : @xmath147 for the macroscopic dynamics , eqn .
( [ x - fin ] ) , we find in pair approximation : @xmath148 \end{aligned}\ ] ] note that the @xmath149 can be expressed in terms of @xmath138 by means of eqn .
( [ eq : cond_prob ] ) .
thus , eqn .
( [ macro - pair ] ) now depends on only two variables , @xmath122 and @xmath138 . in order to derive a _ closed _ form description ,
we need an additional equation for @xmath150 . that can be obtained from eqn .
( [ eq : c - prob ] ) : @xmath151 eqn .
( [ eq : local1 ] ) also requires the time derivative of the global doublet frequency @xmath152 . even in their lengthy form
, the three equations for @xmath153 , @xmath138 , @xmath154 can easily be solved numerically .
this gives the approach some computational advantage compared to averaging over a number of microscopic computer simulations for all possible parameter sets .
although the approach derived so far is quite general in that it can be applied to different network topologies and neighborhood sizes , specific expressions for these three equations of course depend on these .
therefore , in the appendix , these three equations are explicitly derived for a 2d regular lattice with neighborhood @xmath0 using the specific transition rates introduced in the next section . in sect .
[ 4.1 ] , we further show that the pair approximation yields some characteristic quantities such as @xmath155 for the 2d regular lattice in very good agreement with the results of computer simulations .
so far , we have developed a stochastic framework for ( but not restricted to ) nonlinear voter models in a general way , without specifying two of the most important features , namely ( i ) the network topology which defines the neighborhood , and ( ii ) the nonlinearity @xmath9 which defines the response to the local frequencies of the two different states . for ( i ) , let us choose a regular network with @xmath0 , i.e. each voter has 4 different neighbors .
we note explicitly that our modeling framework and the general results derived hold for _ all homogeneous networks _ , but for the visualization of the results it will be most convenient to choose a regular square lattice , where the neighbors appear next to a node .
this allows us to observe the formation of macroscopic ordered states in a more convenient way , without restricting the general case . eventually , to illustrate the dynamics let us now assume a population biology context , where each node is occupied by an individual of either species 0 or 1 .
the spreading of one particular state is then interpreted as the invasion of that respective species and the local disappearence of the other one , while the emergence of a complete ordered state is seen as the complete invasion or domination of one species together with the extinction of the other one .
keeping in mind that we also consider the state of node @xmath24 itself , we can write the possible transition rates , eqn .
( [ eq : nonlinvm ] ) , for the neighborhood of @xmath156 and @xmath157 in the following explicit way ( cf also @xcite ) : @xmath158 eqn .
( [ trans2 ] ) means that a particular node @xmath24 currently in state @xmath157 , or occupied by an individual of species @xmath34 where @xmath34 is either 0 or 1 , will be occupied by an individual of species @xmath159 with a rate @xmath54 that changes with the local frequency @xmath53 in a _ nonlinear _ manner .
the different values of @xmath160 denote the products @xmath161 for the specific values of @xmath88 given .
i.e. , the @xmath160 define the piecewise linear functions shown in fig .
[ fig : nonlin ] .
the general case of six independent transition rates @xmath160 ( n0, ... ,5 ) in eqn . ( [ trans2 ] ) can be reduced to three transition rates @xmath162 , @xmath10 , @xmath11 by assuming a symmetry of the invasion dynamics of the two species , i.e. @xmath163 , @xmath164 and @xmath165 .
further , assuming a pure frequency dependent process , we have to consequently choose @xmath166 , because in a complete homogeneous neighborhood , there is no incentive to change to another state ( there are no other species around to invade ) .
we recall that if the transition rates @xmath10 , @xmath11 are directly proportional to @xmath167 , i.e. @xmath168 and @xmath169 , this recovers the linear vm , eqn .
( [ eq : linvm ] ) .
( note that without the resistence of node @xmath24 discussed in sect .
[ sec : vm ] the linear voter point would read as @xmath170 and @xmath171 instead . ) dependent on the relation of the two essential parameters @xmath10 , @xmath11 , we also find different versions of _ nonlinear _ vm , which have their specific meaning in a population biology context : @xmath172 note , that the parameters @xmath173 can be ordered in @xmath174 different ways .
these reduce to @xmath175 inequalities under the conditions @xmath176 and @xmath177 . in eqn .
( [ eps - alpha ] ) , ( pf ) means ( pure ) _ positive freqency dependent invasion _
, where the transition rate _ increases _ with an increasing number of individuals of the _ opposite _ species @xmath159 in the neighborhood , and ( nf ) means ( pure ) _ negative freqency dependent invasion _ because the transition rate _
decreases_. the two other cases describe positive ( pa ) and negative ( na ) _ allee effects _
these regions are described by @xmath178 inequalities each , all of which show the same relative change in parameter values , if going from @xmath179 to @xmath180 .
similar to the drawings in fig .
[ fig : nonlin ] this can be roughly visualized as an up - down - up change in region ( pa ) and a down - up - down change in region ( na ) . the different parameter regions are shown in fig .
[ region ] . on a first glimpse
, one would expect that the dynamics as well as the evolution of global variables may be different in these regions .
thus , one of the aims of this paper is to investigate whether or to what extent this would be the case .
( 1,1 ) ( 0,0.5)(0.5,0.5)(1,1)(0,1)(0,0.5 ) ( 0,0)(0.5,0.5)(1,0.5)(1,0)(0,0 ) ( 0.,0.)(0.,0.5)(0.5,0.5)(0,0 . ) ( 0.5,0.5)(1,0.5)(1,1)(0.5,0.5 ) ( 0,0)(0,1)(1,1)(1,0)(0,0 ) ( 0,0)(1,1 ) ( 0.2,0.4)@xmath181 ( 0.13,0.30)*(pf ) * ( 0.85,0.65)*(nf ) * ( 0.3,0.75)*(pa ) * ( 0.7,0.2)*(na ) * ( 0.5,-0.2)@xmath10 ( -0.2,0.5)@xmath11 in order to find out about the influence of the nonlinear response function @xmath9 , which is specified here in terms of @xmath10 , @xmath11 , let us start with the mean - field approach that lead to eqn .
( [ eq : master_macro_mean ] ) . as we outlined in sect .
[ 3.2 ] , the calculation of the @xmath182 in eqn .
( [ eq : master_macro_mean ] ) requires to look at each possible occupation pattern @xmath28 for a neighborhood @xmath25 , for instance , @xmath183 .
the mean - field approach assumes that the occurence of each 1 or 0 in the pattern can be described by the global frequencies @xmath20 and @xmath184 , respectively ( for simplicity , the abbreviation @xmath185 will be used in the following ) . for the example of string @xmath183
we find @xmath186 . the same result yields for @xmath187 and for any other string that contains the same number of 1 and 0 , i.e. there are exactly @xmath188 different possibilities . for strings with two nodes of each species , @xmath189 times the contribution @xmath190 results , etc . inserting eqn .
( [ trans2 ] ) for the transition rates , we find with @xmath166 the equation for the mean - field dynamics : @xmath191 \\ \ ; - x & \big [ ( 1-\alpha_{1 } ) ( 1-x)^{4 } + 4 ( 1-\alpha_{1 } ) x(1-x)^{3 } \\ & + 6 \alpha_{2 } x^{2}(1-x)^{2 } + 4 \alpha_{1 } x^{3 } ( 1-x ) \big ] \end{aligned}\ ] ] the fixed points of the mean - field dynamics can be calculated from eqn .
( [ eq : mean_field ] ) using @xmath192 .
we find : @xmath193 the first three stationary solutions denote either a complete invasion of one species or an equal share of both of them .
`` nontrivial '' solutions , i.e. a _ coexistence _ of both species with different shares of the total population , can only result from @xmath194 , provided that the solutions are ( i ) real and ( ii ) in the interval @xmath195 .
the first requirement means that the two functions @xmath196 , @xmath197 are either both positive or both negative .
the second requirement additionally results in @xmath198 if @xmath199 and @xmath200 if @xmath201 .
this leads to the phase diagram of the mean - field case shown in fig .
[ phase ] .
( 1,1 ) ( 0,0)(0.2,0)(0.2,0.4)(0,0.3)(0,0 ) ( 0.2,1)(0.2,0.4)(1,0.8)(1,1)(0.2,1 ) ( 0.2,0.4)(0.4667,0)(1,0)(1,0.8)(0.2,0.4 ) ( 0.2,0.4)(0,0.3)(0,0.7)(0.2,0.4 ) ( 0,0.7)(0.4667,0.0 ) ( 0,0.3)(1,0.8 ) ( 0.2,0)(0.2,1 ) ( 0,0)(1,1 ) ( 0.2,0.4)@xmath181 ( 0.1,0.1)*a * ( 0.075,0.45)*b * ( 0.1,0.9)*c * ( 0.5,0.9)*d * ( 0.7,0.3)*e * ( 0.3,0.1)*f * ( 0.5,-0.2)@xmath10 ( -0.2,0.5)@xmath11 ( 0.68,0.72)@xmath202 ( 0.55,0.1)@xmath203 and @xmath11 .
( top ) @xmath204 , ( bottom ) @xmath205 .
the solid lines refer to stable solutions , the dashed lines to unstable ones .
the notations a - f refer to the respective areas in the phase diagram , fig .
[ phase ] .
[ a1a2],title="fig:",width=245 ] and @xmath11 .
( top ) @xmath204 , ( bottom ) @xmath205 .
the solid lines refer to stable solutions , the dashed lines to unstable ones .
the notations a - f refer to the respective areas in the phase diagram , fig .
[ phase ] .
[ a1a2],title="fig:",width=245 ] in order to verify the stability of the solutions , we have further done a perturbation analysis ( see also sect . [
sec : pertub ] ) .
the results can be summarized as follows : * in the regions @xmath206 and @xmath207 of the mean - field phase diagram , fig .
[ phase ] , @xmath208 and @xmath209 are the only stable fixed points of the dynamics , while @xmath210 is an unstable fixed point ( cf . also fig .
[ a1a2]top ) .
species 1 with @xmath211 will most likely become extinct , while it will remain as the only survivor for @xmath212 .
thus , the region @xmath213 can be characterized as the region of _ invasion_. * in region @xmath214 , the mean - field limit predicts the three stable fixed points @xmath113 , @xmath60 and @xmath215 . the attractor basin for @xmath215 is the largest as fig .
[ a1a2](top ) indicates .
the separatices are given by the unstable solutions @xmath194 , eqn .
( [ zero ] ) . in this parameter region , the mean - field limit predicts either _ coexistence _ of both species with equal shares , or _
invasion _ of one species , dependent on the initial condition @xmath216 .
* in the regions @xmath217 and @xmath218 , only one stable fixed point @xmath219 can be found , while the solutions @xmath113 and @xmath60 are unstable ( cf . also fig .
[ a1a2]bottom ) .
thus , the mean - field approach predicts the _ coexistence _ of both species with equal share . * finally , in region @xmath88 the solutions 0 , 1 and @xmath215 are unstable fixed points , but the two remaining solutions @xmath194 , eqn .
( [ zero ] ) are stable fixed points ( cf .
[ a1a2]bottom ) .
thus , this region is the most interesting one , since it seems to enable `` nontrivial '' solutions , i.e. an _ asymmetric coexistence _ of both species with different shares .
we note again , that this is a prediction of the mean - field analysis . at the intersection of regions @xmath88 and @xmath206 , these two solutions approach @xmath113 and @xmath60 , while at the intersection of regions @xmath88 and @xmath218 they both converge to @xmath215
. we will compare these mean - field predictions both with computer simulations and analytical results from the pair approximations later in this paper .
before , in sects .
[ sec : limit ] , [ sec : pertub ] we would like to point to some interesting @xmath16 combinations in this phase diagram where the mean - field analysis does not give a clear picture of the dynamics . the first set of interesting points are @xmath16 combinations of values 0 and 1 , such as @xmath220 etc .
these cases are special in the sense that they describe the _ deterministic _ limit of the nonlinear voter dynamics . whereas for @xmath221
always a finite probability exist to change to the opposite state , for @xmath222 the state of node @xmath24 _ never _ changes as long as at least half of the nearest neighbor nodes are occupied by the same species .
on the other hand , it will _ always _ change if more then half of the neighboring nodes are occupied by the _ other _ species .
this refers to a _ deterministic _ positive frequency invasion process .
similarly , a _
deterministic _ negative frequency invasion process is described by @xmath223 .
the deterministic dynamics , as we know from various other examples , may lead to a completely different outcome as the stochastic counterpart . in order to verify that we have conducted computer simulations using a _ cellular automaton _ ( ca ) , i.e. , a two - dimensional regular lattice with periodic boundary conditions and _ synchronous update _ of the nodes
. the latter one can be argued , but we verified that there are no changes in the results of the computer simulations if the sequential update is used . the time scale for the synchronous update is defined by the number of simulation steps . if not stated otherwise , the initial configuration is taken to be a random distribution ( within reasonable limits ) of both species , i.e. initially each node is randomly assigned one of the possible states , @xmath195 .
thus , the initial global frequency is @xmath224 .
[ determ ] shows snapshots of computer simulations of the deterministic dynamics taken in the ( quasi-)stationary dynamic regime .
if we compare the snapshots of the _ deterministic _ voter dynamics with the mean - field prediction , the following observations can be made : 1 .
a spatial coexistence of both species is observed for the @xmath16 values @xmath222 , fig .
[ det00 ] , @xmath225 , fig .
[ det10 ] , @xmath223 , fig .
[ det11 ] , where the global frequency in the stationary state is @xmath219 .
this contradicts with the mean - field prediction for @xmath222 , which is part of region ( a ) and thus should display complete invasion .
a complete invasion of one species is observed for @xmath226 , fig .
[ det01 ] .
this would agree with the mean - field prediction of _ either _ coexistence _ or _ invasion
. a closer look at the bifurcation diagram , fig .
[ a1a2](top ) , however tells us that for the given intial condition @xmath227 the stationary outcome should be _ coexistence _ , whereas the deterministic limit shows _ always invasion _ as was verified by numerous computer simulations with varying initial conditions .
3 . for the deterministic frequency
dependent processes @xmath222 , @xmath225 the spatial pattern becomes stationary after a short time .
for the negative frequency dependence @xmath223 the pattern flips between two different configurations at every time step .
so , despite a constant global frequency @xmath219 in the latter case , local reconfigurations prevent the pattern from reaching a completely stationary state , but it may be regarded as ( quasi-)stationary , i.e. the macroscopic observables do not change but microscopic changes still occur .
4 . in both
positive and negative deterministic frequency dependence cases , individuals of the same species tend to aggregate in space , albeit in different local patterns . for the positive frequency dependence
, we see the occurence of small clusters , fig .
[ det00 ] , or even complete invasion , fig .
[ det01 ] , based on the local feedback within the same species . for the negative frequency dependence however we observe the formation of a meander - like structure that is also known from physico - chemical structure formation @xcite .
it results from the antagonistic effort of each species to avoid individuals of the same kind , when being surrounded by a majority of individuals of the opposite species . in conclusion ,
the mean - field analysis given in this section may provide a first indication of how the nonlinearities may influence the voter dynamics .
this , however , can not fully extended to the limiting cases given by the deterministic dynamics .
the other interesting @xmath16 combination is the linear voter point @xmath228 where _ all _ different regions of the mean - field phase diagram , fig .
[ phase ] , intersect . inserting @xmath228 into eqn .
( [ eq : mean_field ] ) yields @xmath229 regardless of the value of @xmath20 , i.e. @xmath230 for _ all _ initial conditions .
this important feature of the linear vm was already discussed in sect .
[ sec : vm ] .
we recall that , while on the one hand the microscopic realizations always reach consensus ( complete invasion of one species ) in the long term , on the other hand an averaged outcome over many realizations shows that the share of the winning species is distributed as @xmath59 . to put it the other way round , the mean - field limit discussed failes here because it does not give us any indication of the fact that there is a completely ordered state in the linear vm .
the averaged outcome , for example @xmath210 , can result both from complete _ invasion _ of species 1 ( 50__coexistence _ _ of the two species .
both of these outcomes exist in the immediate neighborhood of the linear voter point as fig .
[ phase ] shows . in order to get more insight into this
, we will later use the pair approximation derived in sect . [
3.2 ] . here ,
we first follow a perturbation approach , i.e. we add a small perturbation to the solution describing the macroscopic ordered state of complete invasion ( consensus ) . in terms of the nonlinear response function @xmath9 , expressed by the @xmath160 in eqn .
( [ trans2 ] ) , this means a nonzero value of @xmath231 , i.e. a small parameter indicating the perturbation . with this
, we arrive at a modified mean - field equation : @xmath232 + \frac{dx}{dt}\ ] ] where @xmath233 is given by the nonperturbated mean - field eqn .
( [ eq : mean_field ] ) and the index @xmath234 shall indicate the presence of the perturbation @xmath235 .
consequently , this changes both the value of the fixed points , previously given by eqn .
( [ zero ] ) and their stability . instead of a complete analysis in the @xmath236 parameter state
, we restrict the investigations to the vicinity of the linear voter point @xmath237 where @xmath229 .
( [ eq : mean_field - p ] ) then returns only one real stationary solution , @xmath238 , which is independent of @xmath235 and stable .
consequently , any finite perturbation will destroy the characteristic feature of reaching an ordered state in the linear vm , i.e. complete invasion , and leaves only _ coexistence _ of both species as a possible outcome .
this is little surpising because adding an @xmath239 to the dynamics transforms the former attractor @xmath240 into a repellor , i.e. it prevents reaching the ordered state .
more interesting the question is , how the perturbated linear voter dynamics looks in detail .
this is investigated in the next section by means of computer simulation , and in sect . [ 4.1 ] by means of the pair approximation approach . for further insight into the dynamics of the nonlinear vm , we perform some computer simulations using the ca approach already described in sect .
[ sec : limit ] .
is important to notice that we have chosen different sets of the parameters @xmath10 , @xmath11 from the region of _ positive frequency dependence _
, as defined in fig .
[ region ] .
i.e. , the transition towards the opposite state strictly increases with the number of neighbors in that state ( _ majority voting _ ) .
so one would naively expect a similar macroscopic dynamics in that region as done in @xcite this however is not the case as the following simulations indicate .
a thorough analysis is presented in sect . [ 4.3 ] in order to study the stability of the global dynamics for the different @xmath10 , @xmath11 settings in the vicinity of the linear voter point , we have added a small perturbation @xmath231 . as the investigations of sect .
[ sec : pertub ] have indicated , we should no longer expect consensus for the perturbated linear vm , but some sort of coexistence .
in fact , we observe an interesting nonstationary pattern formation we call _ correlated coexistence_. fig .
[ posit ] ( obtained for another range of parameters @xmath10 , @xmath11 ) shows an example of this .
we find a long - term coexistence of both species , which is accompanied by a spatial structure formation . here
, the spatial pattern remains always nonstationary and the global frequency randomly fluctuates around a mean value of @xmath241 , as shown by fig .
[ x - t-2 ] .
( a)(b ) ( c)(d ) , @xmath169 , @xmath242 , and ( b ) for the same setup and parameters as in fig .
[ posit ] .
the initial frequency is @xmath243=0)=0.5 for both runs .
[ x - t-2],width=245 ] in more specific terms , the regime defined as correlated coexistence is a paramagnetic phase with finite domain length , typical of partially phase - separated systems .
we mention that such a regime was also observed in some related investigations of the vm and other nonlinear spin models with ising behavior @xcite .
also , a similar transition was observed in the abrams - strogatz model @xcite , where the transition rate is a power @xmath206 of the local field .
the stability of the solutions then changes at @xmath244 , from coexistence for @xmath245 to dominance for @xmath246 . in order to find out about the range of parameters in the nonlinear vm resulting in the quite interesting phenomenon of correlated coexistence , we have varied the parameters @xmath10 , @xmath11 within the region of positive frequency dependence .
[ posit ] actually shows results from a set picked from region ( f ) in fig .
[ phase ] , where the mean - field analysis predicts an asymmetric coexistence of both species .
obviously , the nonstationarity results from the perturbation @xmath235 .
however , for other sets @xmath10 , @xmath11 in the positive frequency dependence region the perturbation does _ not prevent _ the system from reaching a global ordered state , i.e. invasion of one species as fig .
[ stoch ] verifies .
this process is accompanied by a clustering process and eventually a segregation of both species indicated by the formation of spatial domains .
[ x - t ] depicts the evolution of the global frequency @xmath14 of species 1 for different intitial frequencies @xmath216 . in every case
, one species becomes extinct . for @xmath212 species 1
is the most likely survivor , while for @xmath211 it is most likely to become extinct . for @xmath224 , random events during the early stage decide about its outcome . on the other hand
, the perturbation also does _ not induce _ an ordered state as the random coexistence in fig .
[ random ] shows , which was again obtained from parameter settings in the region of positive frequency dependence .
so , we conclude that computer simulations of _ positive _ frequency dependent processes show three different dynamic regimes ( dependent on the parameters @xmath10 , @xmath11 ) : ( i ) complete invasion , ( ii ) random coexistence , and ( iii ) correlated coexistence . while for ( i ) and ( ii ) the outcome is in line with the mean - field prediction shown in fig .
[ a1a2 ] , this does not immediately follows for ( iii ) .
so , we are left with the question whether the interesting phenomenon of correlated coexistence is just _ because _ of the perturbation of some ordered state , or whether it may also exist _ inspite _ of @xmath235 .
( a)(b ) ( c)(d ) .
the initial frequencies @xmath243=0 ) of the four different runs are : ( a ) 0.6 , ( b ) 0.5 , ( c ) 0.5 , ( d ) 0.4.,width=245 ] .
[ x - t ] we just add that for _ negative _ frequency dependent invasion , eqn .
( [ eps - alpha ] ) , the the spatial pattern remains random , similar to fig .
[ random ] .
furthermore , regardless of the initial frequency @xmath216 , on a very short time scales , a global frequency @xmath247=0.5 is always reached .
that means we always find _ coexistence _ between both species .
we conclude that for _ negative _ frequency dependence @xmath247=0.5 is the only stable value , which is in agreement with the mean - field prediction , whereas for _ positive _ frequency dependence the situation is not as clear .
to answer the question what ranges of @xmath10 , @xmath11 eventually lead to what kind of macroscopic dynamics , we now make use of the pair approximation already derived in sect . [ 3.2 ] as a first correction to the mean - field limit . here , we follow a two - step strategy : first , we investigate how well the pair approximation , eqs .
( [ eq : master2d ] ) , ( [ eq : doublet2d ] ) , ( [ eq : master2dc ] ) of the macroscopic dynamics , eqn . ( [ x - fin ] ) , predict the global quantities @xmath122 and @xmath138 . in order to specify the network topology , we use again
the ca described above .
second , we use the pair approximation to derive a phase diagram in the case of local interaction . eventually , we test whether these findings remain stable against perturbations of the ordered state .
all predictions are tested by comparison with computer simulations of the microscopic model , from which we calculate the quantities of interest and average them over 50 runs .
here we have to distinguish between the three different dynamic regimes already indicated in sect . [ 2.4 ] . regime ( i ) , _ complete invasion _ , is characterized by fixed points of the macroscopic dynamics of either @xmath248 or @xmath249 .
the ca simulations as well as also the pair approximation of the dynamics quickly converge to one of these attractors , dependent on the initial conditions . regime ( ii ) , _ random coexistence _
, has only one fixed point , @xmath250 , to which the ca simulations quickly converge .
the pair approximation converges to @xmath251 after some initial deviations from the ca simulation , i.e. it relaxes on a different time scale ( @xmath252 ) , but is correct in the long run .
the approximation of the local correlation @xmath138 shows some deviations from the predicted value , @xmath253 .
we have tested the case of random coexistence for various parameter values and found values for @xmath138 between 0.4 and 0.6 .
the discrepancy is understandable , since in the case of long - term coexistence some of the spatial patterns flip between two different random configurations with high frequency .
thus , while the global frequency settles down to 0.5 , the microscopic dynamics is still nonstationary . regime ( iii ) , _ correlated coexistence
_ , the most interesting one , is chacterized by an average global frequency of @xmath251 again , however the existing local correlations lead to a much higher value of @xmath254 .
this is shown in fig .
[ variat ] , where we find @xmath255 from the ca simulations and @xmath256 from the pair approximation .
i.e. , for the case of spatial domain formation the long - range correlations can be well captured by the pair approximation , whereas this was less satisfactory for the short - range correlations of the random patterns . with min - max deviations ( top ) and spatial correlation @xmath138 ( bottom ) for the case of correlated coexistence .
the two curves shown in the bottom part result from averaging over 50 ca simulations ( black dotted line ) and from the pair approximation ( green solid line ) the parameters are as in figs .
[ posit ] , [ x - t-2 ] .
( top ) [ variat],title="fig:",width=245 ] ) , ( bottom ) pair approximation .
phase boundaries for @xmath257 resulting from @xmath258 ( upper limit ) : left red solid line , right black solid line , for comparison phase boundaries resulting from @xmath259 ( lower limit ) : left red solid line ( identical with @xmath258 ) , right red solid line on the far right . dashed red lines indicate the shift of the phase boundaries for @xmath259 if @xmath260 .
the linear voter point ( 0.2,0.4 ) is indicated by @xmath181 .
further , @xmath261 marks those three parameter sets from the positive frequency dependence region where ca simulations are shown in figs .
[ stoch ] , [ posit ] , [ random ] ( see also global frequency @xmath122 and local correlation @xmath138 in figs .
[ x - t-2 ] , [ x - t ] , [ variat ] ) .
the straight dashed lines mark the different parameter areas given in eqn .
( [ eps - alpha ] ) which are also shown in fig .
[ region].,title="fig:",width=245 ] the insights into the dynamics of the nonlinear vm derived in this paper are now summarized in a phase diagram that identifies the different parameter regions @xmath16 for the possible dynamic regimes identified in the previous sections . in order to find the boundaries between these different regimes , we carried out ca simulations of the complete parameter space , @xmath262 .
precisely , for every single run the long - term stationary values of @xmath20 and @xmath138 were obtained and then averaged over 50 simulations .
as described above , the three different regimes could be clearly separated by their @xmath263 values , which were used to identify the phase boundary between the different regimes .
the outcome of the ca simulations is shown in the phase diagram of fig .
[ phasepl ] ( top ) and should be compared with fig .
[ phase ] , which results from the mean - field analysis in sect .
[ 3.2 ] and thus neglects any kind of correlation . instead of the six regions distinguished in fig .
[ phase ] , in the local case we can distinguish _ two _ different regions divided by one separatrix : the parameter region left of the separatrix refers to the _ complete invasion _ of one of the species , with _
high _ local correlations during the evolution . in the ca simulations ,
we observe the formation of domains that grow in the course of time until exclusive domination prevails .
asymptotically , a _
stationary _ pattern is observed , with @xmath264 ( or 0 ) and @xmath265 ( or 0 ) ( cf the simulation results show in figs .
[ stoch ] , [ x - t ] ) .
i.e. , the system converges into a frozen state with no dynamics at all .
the region to the right of the separatrix refers to _ random coexistence _ of both species with @xmath251 and _ no _ local correlations , i.e. @xmath253 . in the ca simulations , we observe _ nonstationary _ random patterns that change with high frequency ( cf .
[ random ] ) .
both regions are divided by a _
separatrix_. as shown in fig .
[ phasepl ] ( top ) , the separatrix is divided into two pieces by the linear voter point , ( 0.2,0.4 ) . above that point
, the separatrix is very narrow , but below the voter point it has in fact a certain extension in the parameter space .
looking at the dynamics _ on _ the separatrix , we find that the mean frequency is @xmath251 both above and below the voter point ( see also fig .
[ variat ] , top ) .
the local correlation @xmath266 holds within the whole extended area of the separatrix which identifies it as the region of _ correlated coexistence _
( see also fig .
[ variat ] , bottom ) .
the question is how well this phase diagram can be predicted by using the pair approximation of the dynamics , described by eqs .
( [ eq : master2d ] ) , ( [ eq : master2dc ] ) . for a comparison
, the coupled equations were numerically solved to get the asymptotic solutions for @xmath263 ( which looks complicated but is numerically very fast and efficient ) . in order to distinguish between the three regimes , we have to define a critical @xmath138 value for the correlations .
whereas in the ca simulations @xmath266 indicated a correlated nonstationary coexistence , this value was never reached using the pair approximation ( see fig .
[ variat ] , bottom ) , so @xmath267 could be regarded as an _ upper limit _ for the case of correlated coexistence . random coexistence , on the other end , yielded @xmath253 in the ca simulations and a value between 0.4 and 0.6 in the pair approximation .
so , given suitable initial conditions , we can regard @xmath268 as a _
lower limit _ , to identify correlated coexistence .
the results are shown in fig .
[ phasepl ] ( bottom ) which shall be compared with the phase diagram above ( fig .
[ phasepl ] , top ) .
[ phasepl ] ( bottom ) shows the influence of the @xmath138 threshold value . with the lower limit
, we find a quite broad region of correlated coexistence , which for example also includes the point @xmath269 for which a random coexistence in the ca simulations was shown in fig .
[ random ] . using the upper limit @xmath258 results in a much smaller region of correlated coexistence . comparing this with the ca simulations above
, we can verify that the pair approximation correctly predicts the extended region below the voter point and also shows how it becomes more narrow above the voter point .
one should note that the left border of the separatrics is not affected by the threshold value , whereas the right border shifts considerably .
the left border also contains the voter point ( independent of the @xmath138 threshold value ) , for which a complete invasion can be observed .
therefore , it is quite interesting to look into changes of the phase diagram if additional perturbations are considered ( see also sect .
[ sec : pertub ] ) .
[ phasepl ] ( bottom ) shows ( for the threshold @xmath270 ) that this does _ not _ affects the existence of the three dynamic regimes and most notably of the extended separatrix below the voter point , but only shifts the boundaries toward the left , dependent on the value of @xmath231 ( this can be also verified for @xmath258 but is omitted here , to keep the figure readable ) .
thus , the consideration of perturbations in the phase diagram reveals that it is indeed the _
nonlinearity _ in the voter model which allows for the interesting phenomenon of the correlated coexistence and _ not _ just the perturbation . a closer look into fig .
[ phasepl ] ( bottom ) also shows that in the perturbated phase diagram the voter point no longer lies on the boundary towards the region of complete invasion but clearly _ within _ the region of correlated coexististence .
this is in agreement with the findings in sect .
[ sec : pertub ] which showed that for the linear vm complete invasion is an unstable phenomenon and changes into correlated coexistence for finite @xmath235 .
in this paper , we investigated a local model of frequency dependent processes , which for example models the dynamics of two species @xmath195 in a spatial environment .
individuals of these species ( also called voters ) are seen as nodes of a network assumed as homogeneous in this paper ( i.e. all nodes have the same number of neighbors , @xmath25 ) . the basic assumption for the microscopic dynamics
is that the probability to occupy a given node with either species @xmath113 or @xmath60 depends on the frequency of this species in the immediate neighborhood .
different from other investigations , we have counted in the state of the center node as well ( see sect . [ sec : vm ] ) and have further considered a _ nonlinear response _ of the voters to the local frequencies .
studies of a nonlinear version of the traditional voter model ( without counting the state of the central node and with sequential dynamics ) have already been analyzed before .
@xcite , as pointed out before , is closest to our investigations , but restricted itself to the mean - field analysis and computer simulations of the 2d case , to obtain a phase diagram similar to fig .
[ phasepl ] ( top ) .
@xcite , on the other hand , have provided a markov analysis which is restricted only to very small ca .
the two - parameter model in @xcite is for @xmath271 a nonlinear voter model that exhibits at the voter point ( @xmath272 ) a transition from a ferromagnetic phase , i.e. invasion , for @xmath273 to a paramagnetic phase ( correlated coexistence ) for @xmath274 . also
the case of the perturbated linear voter model is included in the model , for @xmath272 and @xmath275 .
similar results are also presented in @xcite , which points out relations to random branching processes , and in @xcite , where the emphasis is on investigations of the interface density , to describe the coarsening process .
a recent paper @xcite also shows for spin systems with two symmetric absorbing states ( such as the vm ) that the macroscopic dynamics only depends on the first derivatives of the spin - flip probabilities . in our paper
, we set out for a formal approach that allows to derive the dynamics on different levels : ( i ) a _ stochastic dynamics _ on the microlevel , which is used for reference computer simulations but also allows a derivation of the ( ii ) _ macroscopic dynamics _ for the key variables , given in terms of differential equations .
this macroscopic dynamics is then analysed by two different approximations , ( i ) a _ mean - field approximation _ that neglects any local interaction in the network , and ( ii ) a _ local approximation _ considering correlations between pairs of nearest neighbors . in order to test the validity of these approximations
, we compare their preditions with the averaged outcome of the microscopic computer simulations .
we like to emphasize that our approach is general enough to be applied to various forms of _ frequency dependent processes _ on homogeneous networks with different number of neighbors .
even if a two - dimensional regular lattice is used to illustrate the dynamics , the approach is not restricted to that .
our main result , in addition to the general framework for nonlinear frequency dependent processes , is the derivation of a _ phase diagram _ using the _ pair approximation _ derived in this paper .
this approach predicts correctly both the type of the dynamics and the asymptotic values of the key variables dependent on the possible nonlinearities for the case of _ local interaction _ , @xmath0 .
the predicted phase diagram was verified by comparision with extensive microscopic computer simulations rastering the whole parameter space .
while the structure of the phase diagram was already known from previous computer simulations presented in @xcite we could demonstrate that the pair approximation works very well both for predicting the correct phase boundaries and the dynamics within these phases .
it should be noticed that the pair approximation is a valuable tool , particularly with respect to computational efforts .
the computer simulations are much more timeconsuming , since the results of the different runs have to be averaged afterwards .
the pair approximation , on the other hand , is based on only 2 coupled equations and therefore needs less computational effort . in the following ,
we discuss some of the interesting findings .
_ the region of correlated coexistence : _ analysing the nonlinear vm with _ local interaction _ has shown that there are in fact only _ three _ different dynamic regimes dependent on the nonlinearities @xmath16 : ( i ) complete invasion , ( ii ) random coexistence , and ( iii ) correlated coexistence . the first one is already known as the standard behavior of the _ linear _ vm .
consequently the only interesting feature , namely the time to reach the ordered state dependent on the network size and topology , has been the subject of many investigations @xcite .
number ( ii ) , on the other hand , only leads to trivial results as no real dynamics is observed .
thus , the most interesting regime is ( iii ) correlated coexistence , which can be found in a small , but not negligible parameter region below the voter point .
this region separates the two dominant regimes ( i ) and ( ii ) and therefore was called a separatrix here . going over from the right to the left side of the phase diagram in that region
, we notice a transition from 0.5 to 1.0 ( or 0.0 respectively ) in the mean frequency , and from 0.5 to 0.7 to 1.0 in the local correlations .
thus , in fact @xmath138 separates the two dynamic regimes ( i ) and ( ii ) ( below the voter point ) . for parameters chosen from that region
, we find in the ca simulations a long - term and nonstationary _ coexistence _ between the two species as on the _ right _ side of the phase diagram .
but we also find the long - range spatial correlations that lead to the formation of spatial domains as shown e.g. in fig .
[ posit ] which is characteristic for the _ left _ side of the phase diagram .
the spatial pattern formation is also indicated by large fluctuations of @xmath122 shown in fig . [
variat](top ) . a single run , as shown in fig .
[ x - t-2 ] , clearly indicates the long - term nonstationary coexistence of both species .
we emphasize that the separatrix between the two dynamic regimes is well predicted by the macroscopic dynamics resulting from the pair approximation ( as can be clearly seen in fig .
[ phasepl ] ) .
most importantly , we could verify that the correlated coexistence of both species is not simply the effect of an additional perturbation , but results from the nonlinear interaction .
_ comparison with the mean - field phase diagram : _ in our paper , the mean - field approximation plays the role of a reference state used to demonstrate the differences of the local analysis . the phase diagram of fig .
[ phase ] distinguishes between six different regions , whereas the one in the local case , fig .
[ phasepl ] ( top ) shows only three . comparing the two phase diagrams
, we realize that the most interesting regions in fig .
[ phase ] , namely ( c ) and ( f ) , have simply collapsed into the separatrix shown in fig .
[ phasepl ] .
the region ( c ) of unstable asymmetric coexistence or multiple outcome , respectively ( see sect . [ 3 ] ) , relates to the separatrix line above the voter point .
it should be noticed that the phase diagram for local interaction , fig .
[ phasepl ] , correctly predicts that the deterministic behavior for @xmath226 leads to complete invasion ( see sect . [
4.3 ] and fig .
[ det01 ] ) .
the region ( f ) of stable asymmetric coexistence relates to the extended area of the separatrix shown below the voter point in fig .
[ phasepl ] , where we still see a coexistence of both species - but the asymmetry between the two species relates to their changing dominance over time , as figs .
[ x - t-2 ] , [ variat ] ( top ) clearly illustrate .
we conclude that in the local case no regions of stationary _ and _ asymmetric coexistence between the two species exist , as was predicted by the mean - field analysis .
however , we find a ( small but extended ) region _ on _ the separatrix that shows the _ nonstationary _ and asymmetric coexistence of the two species for _ single _ realizations ( which results in a symmetric coexistence averaged over runs , @xmath251 , see fig .
[ variat ] , top ) . _ the role of positive frequency dependence : _ the possible nonlinear responses in frequency dependent processes can be distinguished in four parameter areas of positive and negative frequency dependence and positive and negative allee effects , as fig .
[ region ] shows .
previous investigations @xcite assigned a dynamic leading to invasion to a positive frequency dependence , while associating a spatial coexistence with negative frequency dependence .
our investigations have shown that such an assignment does not unambiguously hold .
in particular , a random coexistence can be found for _ negative _ frequency dependent dynamics as well as for _ positive _ frequency depencence , which was so far assigned to complete invasion only @xcite . on the other hand ,
complete invasion is not observed only for positive frequency dependence , but also for positive and negative allee effects .
a random spatial coexistence can be found for positive and negative allee dynamics as well .
the only case where just one dynamic regime can be observed is the case of negative frequency dependence .
we note , however , that the nonstationary long - term coexistence with spatial pattern formation occurs both for the positive frequency dependence and the negative allee dynamics , given that the parameters are chosen from the most interesting zone of the separatrix below the voter point . in conclusion ,
the region of positive frequency dependence bears in fact a much more richer dynamics , as it is transected by the separatrix we identified in the local analysis and thus shows all three types of dynamics we could identify for nonlinear voter models , namely ( i ) complete invasion of one of the species via the formation of large domains , ( ii ) long - term coexistence of both species with random distribution , ( iii ) long - term coexistence of both species with formation of nonstationary domains . however , the most interesting regime ( iii ) is _ not _ restricted to positive frequency dependent processes , but can be also found for some negative allee effects .
we summarize our findings by pointing out that _ nonlinear _ vm show indeed a very rich dynamics which was not much investigated yet .
in addition to the phenomenon of complete invasion ( or consensus ) which occurs also beyond the linear vm , we find most interesting that certain parameter settings lead to a dynamics with nonstationarity and long - term correlations .
thinking about possible applications of the vm , we see that in particular this region has the potential to model relevant observations , be it the temporal dominance of certain species in a habitat or the temporal prevalence of certain opions ( or political parties ) in a social system .
the nonstationarity observed gives rise to the prediction that such dominance may not be the end , and change happens ( even without additional perturbation ) .
the authors want to thank thilo mahnig and heinz mhlenbein for discussions on an early version of this paper .
here we derive some explicit expressions for the three equations of the pair approximation discussed in sect . [ 3.2 ] , for the global frequency @xmath122 ( eqn .
( [ macro - pair ] ) ) , the doublet frequency @xmath152 and the correlation term @xmath138 , eqn .
( [ eq : local1 ] ) ) .
the equations are derived for the neighborhood @xmath0 .
we use the notation @xmath276 . using eqn .
( [ eq : cond_prob ] ) and the transition rates of eqn .
( [ trans2 ] ) , we find for @xmath122 , eqn .
( [ macro - pair ] ) in pair approximation : @xmath277 \\ & + 4\alpha_1 \left [ \frac{x}{(1-x)^3 } ( 1 - 2x+xc_{1|1})^3 ( 1-c_{1|1 } ) \right .
\\ & \qquad \quad - x ( 1-c_{1|1 } ) { c_{1|1}}^3 \big ] \\ & + 6 \alpha_2 \left [ \frac{x^2}{(1-x)^3 } ( 1 - 2x+xc_{1|1})^2 ( 1-c_{1|1})^2 \right . \\ & \qquad \quad - x ( 1-c_{1|1})^2 { c_{1|1}}^2 \big ] \\ & + ( 1-\alpha_1 ) \left [ \frac{x^4}{(1-x)^3 } ( 1-c_{1|1})^4 - x ( 1-c_{1|1})^4 \right ] \\ & + 4(1-\alpha_2)\left [ \frac{x^3}{(1-x)^3}(1 - 2x+xc_{1|1 } ) ( 1-c_{1|1})^3 \right .
\\ & \qquad \quad - x ( 1-c_{1|1})^3 c_{1|1 } \big ] \end{aligned}\ ] ] we note that @xmath278 and @xmath279 in the mean - field limit , in which case eqn .
( [ eq : master2d ] ) reduces to eqn .
( [ eq : mean_field ] ) . in order to calculate the time derivative of the doublet frequency @xmath280 we have to consider how it is affected by changes of @xmath34 in a specific occupation pattern of size @xmath0 , @xmath281 , considering the @xmath282 as constant .
again , in a frequency dependent process it is assumed that the transition does not depend on the exact distribution of the @xmath282 , but only on the frequency of a particular state @xmath34 in the neighborhood .
let @xmath283 describe a neighborhood where the center node in state @xmath34 is surrounded by @xmath284 nodes of the same state @xmath34 .
for any given @xmath285 , there are @xmath286 such occupation patterns .
the global frequency of neighborhood @xmath283 is denoted as @xmath287 with the expectation value @xmath288 .
obviously , @xmath287 can be calculated from the global frequencies @xmath289 of all possible occupation distributions @xmath120 ( eqn .
( [ sigm01 ] ) ) , that match the condition @xmath290 i.e. it is defined as @xmath291 regarding the possible transitions , we are only interested in changes of the doublet ( 1,1 ) , i.e. transitions @xmath292 or @xmath293 . the transition rates shall be denoted as @xmath294 and @xmath295 respectively , which of course depend on the local neighborhood @xmath283 . with this , the dynamics of the expected doublet frequency can be described by the rate equation : @xmath296 \end{aligned}\ ] ] in order to specify the transition rates of the doublets @xmath297 , with @xmath298 and @xmath299 , we note that there are only 10 distinct configurations of the neighborhood .
let us take the example @xmath300 .
a transition @xmath301 of the center node would lead to the extinction of 4 doublets @xmath302 . on the other hand , the transition rate of the center node is @xmath235 as known from eqn .
( [ trans2 ] ) .
this would result in @xmath303 .
however , for a lattice of size @xmath1 the number of doublets is @xmath304 , whereas there are exactly @xmath1 neighborhoods @xmath32 .
therefore , if we apply the transition rates of the single nodes , eqn .
( [ trans2 ] ) , to the transition of the doublets , their rates have to be scaled by 2 .
similarly , if we take the example @xmath305 , a transition of the center node @xmath306 would occur at the rate @xmath307 and would create 3 new doublets . applying the scaling factor of 2
, we verify that @xmath308 .
this way we can determine the other possible transition rates : @xmath309 note that two of the transition rates are zero , because the respective doublets ( 1,1 ) or ( 0,1 ) do not exist in the assumed neighborhood .
finally , we express the @xmath288 in eqn .
( [ d_frequency2 ] ) by the @xmath310 of eqn .
( [ s - m ] ) and apply the pair approximation , eqn .
( [ eq : pairapprox ] ) , to the latter one . this way
, we arrive at the dynamic equation for @xmath152 : @xmath311 \\ & + 6 \alpha_2 \left[\frac{x^2}{(1-x)^3 } ( 1-x+xc_{1|1})^2 ( 1-c_{1|1})^2 c_{1|1}^{2 } \right.\\ & \qquad \quad -x ( 1-c_{1|1})^2 \big ] \\ & + 2(1-\alpha_1 ) \left [ \frac{x^4}{(1-x)^3 } ( 1-c_{1|1})^4 \right .
\\ & \qquad \quad -x ( 1-c_{1|1})^3 \big ] \\ & + 2(1-\alpha_2 ) \left [ \frac{x^3}{(1-x)^3}3 ( 1 - 2x+xc_{1|1})\times \right .
\\ & \qquad \quad \times ( 1-c_{1|1})^3 - x ( 1-c_{1|1})^3 c_{1|1 } \big ] \end{aligned}\ ] ] the third equation , eqn .
( [ eq : local1 ] ) , for the correlation term @xmath138 can be obtained in explicte form by using eqn .
( [ eq : master2d ] ) for @xmath122 and eqn .
( [ eq : doublet2d ] ) for @xmath312 : @xmath313\\ & + 6\alpha_2 ( 1-c_{1|1})^3 \left[\frac{x}{(1-x)^3 } ( 1-x+xc_{1|1})^2 -c_{1|1}\right]\\ & + ( 1-\alpha_1)(1-c_{1|1})^4 \left[\frac{x^3}{(1-x)^3 } ( 2-c_{1|1 } ) + c_{1|1}\right ] \\ & + ( 1-\alpha_2)(1-c_{1|1})^3 \big[2c_{1|1}(2c_{1|1}-1 ) \\ & \quad + \frac{x^2}{(1-x)^3}(1 - 2x+xc_{1|1})(6 - 4c_{1|1 } ) \big ] \end{aligned}\ ] ] kendall , b. e. ; bjrnstad , o. n. ; bascompte , j. ; keitt , t. h. ; fagan , w. f. ( 2000 ) .
dispersal , environmental correlation , and spatial synchrony in population dynamics .
_ the american naturalist _ * 155(5 ) * , 628636 .
oomes , n. a. ( 2002 ) .
emerging markets and persistent inequality in a nonlinear voter model . in : d. griffeath ; c. moore ( eds . ) , _ new constructions in cellular automata _ , oxford universitiy press
. 207230 . | in nonlinear voter models the transitions between two states depend in a nonlinear manner on the frequencies of these states in the neighborhood .
we investigate the role of these nonlinearities on the global outcome of the dynamics for a homogeneous network where each node is connected to @xmath0 neighbors .
the paper unfolds in two directions .
we first develop a general stochastic framework for frequency dependent processes from which we derive the macroscopic dynamics for key variables , such as global frequencies and correlations .
explicit expressions for both the mean - field limit and the pair approximation are obtained .
we then apply these equations to determine a phase diagram in the parameter space that distinguishes between different dynamic regimes .
the pair approximation allows us to identify three regimes for nonlinear voter models : ( i ) complete invasion , ( ii ) random coexistence , and most interestingly
( iii ) correlated coexistence .
these findings are contrasted with predictions from the mean - field phase diagram and are confirmed by extensive computer simulations of the microscopic dynamics .
_ pacs _ : 87.23.cc population dynamics and ecological pattern formation , 87.23.ge dynamics of social systems |
the works on meson baryon dynamics of ref .
@xcite showed how chiral symmetry constraints could be accommodated within a unitarity approach , able to describe resonances .
this proved to be crucial to disentangle the intricate interaction between antikaons and nucleons at low energies @xcite@xcite .
the model of ref .
@xcite , was employed in ref .
@xcite to microscopically derive an optical potential for the @xmath0 in nuclear matter in a self - consistent manner .
self - consistency turned out to be a crucial ingredient to derive the @xmath6-nucleus potential and led to an optical potential considerably more shallow than those found in refs .
@xcite@xcite . in refs .
@xcite and @xcite , the predictions of the chirally inspired potential of ref .
@xcite for measured shifts and widths of @xmath0 atoms were evaluated , and it was found that this potential provides an acceptable description of the observed kaonic atom states , through the whole periodic table . despite of having both real and imaginary parts of quite different depth , some other empirical optical potentials ( @xcite@xcite ) also examined in ref .
@xcite , led to acceptable descriptions of the experimentally available @xmath0 atom data as well .
however , there were appreciable differences among the predicted widths for deeply bound antikaon nuclear states , not detected yet , when different potentials were used .
the aim of this paper is to explore the possibility of differentiating between several @xmath0 nucleus optical potentials by means of the scattering data .
the extrapolation to finite @xmath0 kinetic energies of the potential of ref .
@xcite requires at least the inclusion of the @xmath7-wave part of the @xmath0 selfenergy .
this was performed in ref .
@xcite , and tested for @xmath8 scattering in ref .
however , even after having included @xmath7-wave contributions , one can not expect reliable predictions from the theoretical potential of refs .
@xcite and @xcite , at the lowest energy for which there exist experimental data ( @xmath5 mev for the @xmath0 momentum ) , where @xmath9 and @xmath10 waves contributions are relevant .
besides , as we will show , for this relatively high energy , the impulse approximation works reasonably well , which is a clear indication that these data do not have much information on the details of the @xmath0-nucleus dynamics .
thus , we have also focused our attention at the typical momentum of the @xmath0 after the @xmath11-meson decay ( @xmath12 127 mev ) with the hope that the scattering experience could be performed at da@xmath13ne or at kek or in the future japanese hadron collider ( jhc ) .
we solve the klein gordon equation @xmath14 where @xmath15 is the center of mass ( cm ) @xmath0nucleus energy , @xmath16 and @xmath17 are the finite - size coulomb and optical @xmath0nucleus potentials and @xmath18 the reduced @xmath0-nucleus mass . at large distances and for a cm scattering angle @xmath19 , the @xmath0 wave - function @xmath20 behaves as @xmath21 , with @xmath22 and @xmath23 the standard wave functions for coulomb scattering from a punctual charge @xmath24 and @xmath25 the scattering amplitude , which normalization is determined by its relation to the cm differential elastic cross section @xmath26 .
the integrated cross sections read : @xmath27 \sigma_{\rm t } & = & \frac{2\pi}{q^2 } \sum_l ( 2l+1 ) \left [ 1- \eta_l \cos\left(2 \left ( \sigma_l+\delta_l\right)\right ) \right ] , \label{eq : sr}\\[2 mm ] \sigma_{\rm re}&= & \frac{\pi}{q^2}\sum_l ( 2l+1 ) \left [ 1- \eta_l^2\right ] \label{eq : st}\end{aligned}\ ] ] with @xmath28 the cm @xmath0 momentum and @xmath29 , @xmath30 and @xmath31 the standard coulomb phase shifts , the additional phase shifts due to strong interaction and the inelasticities appearing in the standard partial wave decomposition of @xmath25 ( see ref .
while the elastic ( @xmath32 ) and total ( @xmath33 ) cross sections are infinite , the reaction ( @xmath34 ) cross section is finite because of the short - range of the nuclear interaction .
the @xmath0-nucleus optical potential , @xmath17 , is related to the @xmath0-selfenergy , @xmath35 inside of a nuclear medium , neglecting isovector effects , by @xmath36 where @xmath37 and @xmath38 are the @xmath0 mass and laboratory kinetic energy and @xmath39 is the sum of proton and neutron densities . from the antikaon - selfenergy , as determined by , refs .
@xcite and @xcite , we define the first selfenergy used in this work ( @xmath40 ) .
this selfenergy does not have any free parameters , all the needed input is fixed either from studies of meson - baryon scattering in the vacuum or from previous studies of pion - nucleus dynamics @xcite .
it provides an acceptable ( @xmath41 of 2.9 ) description of the set of 63 shifts and widths of @xmath0 atom levels used in ref .
we have neglected all type of non - localities , since they lead to changes in the results presented here of 3% at most . as in ref .
@xcite , we also construct a modified selfenergy , which we call @xmath42 , by adding to @xmath43 a phenomenological part linear in density . this phenomenological part is determined by a constant @xmath44 which we fix to the value @xmath45 , obtained in ref .
@xcite from a @xmath46-fit to the kaonic atom data .
the new selfenergy reads : @xmath47 the third selfenergy considered in this work is just obtained from the impulse approximation ( ia ) , i.e. , @xmath48 form for the @xmath0 selfenergy , and it neglects all orders higher than the leading one , in the density expansion .
it reads : @xmath49 with @xmath50 , @xmath51 the nucleon mass , @xmath52 the total cm @xmath53 energy and @xmath54 the isoscalar and isovector forward antikaon - nucleon scattering amplitudes , which partial wave decomposition reads : @xmath55 with @xmath56 the total angular momentum . at threshold , @xmath57 @xcite .
the ia leads to extremely poor results for kaonic atoms @xcite,@xcite .
this is a clear indication that higher density corrections , not taken into account within the ia , are extremely important for kaonic atoms .
finally , we have also considered two other antikaon selfenergies fitted to the kaonic atom data and energy independent ( @xcite and @xcite ) : @xmath58 \pi^{\rm iaph } & = & - 4 \pi \left ( 1 + \frac{\mu}{m}\right ) { \tilde b}_0 \rho \label{eq : iaph } \end{aligned}\ ] ] with @xmath59 , @xmath60 and @xmath61 as determined from @xmath46-fits to @xmath0-atom data in ref .
note that , though both @xmath62 and @xmath63 are linear in density selfenergies , they lead to substantially different potentials , since the real parts of the coefficients @xmath64 and @xmath65 differ both in sign and in size .
since the @xmath0 lifetime is relatively small , in practical terms it is experimentally difficult to count with low energetic @xmath0 beams .
however , all selfenergies described in the previous section , except for that obtained in the ia , are valid only near threshold .
thus , we have studied the case @xmath66 mev , since this is the @xmath0 momentum after the @xmath11-meson decay .
mev @xmath0 from @xmath1c , @xmath2ca and @xmath3pb with different @xmath0 selfenergies.,height=559 ] in fig .
[ fig : comp1 ] we present results obtained with the @xmath0 selfenergies fitted to the kaonic atom data .
we also show results obtained by using the ia , where we have approximated the ia selfenergy at @xmath67 mev by its threshold value quoted above .
strong interaction integrated elastic , reaction and total cross sections are also given in the top part of table [ tab : comp1 ] .
we obtain these cross sections after having got rid of the coulombian interaction , otherwise the total and elastic cross sections would diverge , i.e. , we compute strong phase - shifts and inelasticities ( @xmath30 and @xmath31 ) in presence of the coulomb interaction , and afterwards we set to zero the coulombian phase shifts , @xmath29 , in the formulae of eqs .
( [ eq : se])-([eq : st ] ) .
mev @xmath0 from @xmath1c and @xmath2ca .
data are taken from ref .
@xcite.,height=362 ] .
strong integrated elastic , reaction and total cross sections ( in mb ) at @xmath67 ( top ) and 300 ( bottom ) mev . [ cols="<,^,^,^,^,^,^,^,^,^ " , ] as can be seen in the figure , the non - linear density dependent @xmath0 selfenergy , @xmath68 , and the linear density dependent , threshold ia selfenergy , @xmath69 , provide extraordinarily similar results . since both models have the same linear term in density ,
this is a clear indication that the reaction takes place in the surface of the nuclei , because of the big imaginary part of the potentials .
the semiphenomenological @xmath70 selfenergy , has a stronger departure from a linear behaviour in density than @xmath68 , it has a smaller imaginary part ( see fig . 1 of ref
. @xcite ) and all of these explain the bigger differences with the ia model .
results , in particular position of the minima , obtained with @xmath70 are clearly distinguishable from those obtained with any of the other three models also plotted in the figure , pointing out to a clear different density behavior likely due to the selfconsistent derivation of it . as a matter of example , for @xmath1c in the region around @xmath71 , 2dd , ia and iaph give similar elastic cross sections of about 33 mb / sr , whereas thph gives about 52 mb / sr .
this difference is appreciable and the size of the cross sections , tens of mb / sr , might allow to measure such a difference at da@xmath13ne or kek or in the future at the jhc .
the differences are even bigger for larger angles , around the minimum of the thph cross section ( region 110-@xmath72 ) , but there , the cross sections are smaller , which makes harder to get the required statistics to see the effect . besides , theoretical results in the neighborhood of a minimum are subject to more uncertainties .
similar conclusions can be drawn from the @xmath2ca and @xmath3pb results .
in what respects to the integrated cross sections of table [ tab : comp1 ] , 2dd and ia give similar cross sections , though the ia reaction cross section is always slightly bigger , because the imaginary part of the @xmath73 parameter in eq .
( [ eq:2dd ] ) is negative .
the iaph model always provides the biggest reaction cross sections , because its selfenergy has also the largest imaginary part among all models considered ( see fig . 1 of ref
. @xcite ) .
thus one can differentiate two sets of models , i.e , 2dd and ia selfenergies from thph and iaph ones .
besides , measurements , with precisions of about 10% , of the reaction cross sections would disentangle between thph and iaph models .
let us look now to the experimental data .
there only exist data @xcite on @xmath0 differential elastic cross sections for @xmath74 mev and from @xmath1c and @xmath2ca . in fig .
[ fig:800 ] we compare the ia predictions ( solid line ) , including up to @xmath75waves ( from ref .
@xcite ) , to data .
there also exist some data on total cross sections ( mb ) , @xmath76 [ @xmath77 , @xcite from @xmath1c at @xmath74 [ @xmath78 mev .
the ia@xmath79 model provides again an acceptable description : 345 mb at 800 mev and 304 mb at 655 mev .
thus , this region is less sensitive to in nuclear medium effects than the @xmath0 atom one .
indeed , the glauber approximation also describes the 800 mev scattering data , as discussed in @xcite .
this work also corroborates that the imaginary part of the @xmath80 amplitude , obtained from the @xmath0-nucleus scattering data , is close to that deduced in the vacuum . besides
, the contribution of @xmath9 and @xmath10 waves , not included in the thph model , turn out to be important ( compare the solid line to the ia results obtained when only @xmath81 and @xmath7 waves dot - dashed line are considered , in fig .
[ fig:800 ] ) .
in addition , the models of refs .
@xcite and @xcite for the @xmath81 and @xmath7 @xmath80 waves , though realistic near threshold , can not be safely extrapolated to momenta as high as 800 mev .
thus , one expects the poor description of data provided by the thph model .
it is however surprising , that the iaph predictions turn out to be almost indistinguishable from the ia@xmath79 ones .
this is merely a coincidence and , it occurs since accidentally at this momentum , @xmath82 is approximately equal to @xmath64 .
mev @xmath0 from @xmath1c , @xmath2ca and @xmath3pb with different @xmath0 selfenergies.,height=555 ] to finish , we also present results at an intermediate @xmath0 momentum ( @xmath83 mev ) , despite the fact that there exist no data . for this momentum , calculations based on
the ia shows that higher waves than the @xmath7 one have a small / moderate contribution and therefore can be neglected in some approximations .
thus in fig .
[ fig : comp2 ] and bottom part of table [ tab : comp1 ] , we compare again the thph , 2dd and iaph models for the @xmath0 selfenergy inside the nuclear medium , together with the ia results including up to the @xmath7-wave , or up to the @xmath9-wave ( partial waves are taken from ref .
the first observation is that the 2dd model differs now more than for the @xmath4 mev momentum case , from the ia models .
this is mainly due to the effect of @xmath7-wave in the latter ones .
the second observation is that the semiphenomenological model thph leads to a pattern clearly different than the rest of selfenergies , not only for the elastic differential cross section but also in the integrated ones compiled in table [ tab : comp1 ] .
this is in principle good news , because then a scattering measurement in this region of @xmath0 momentum will be definitive to disentangle between this approach and the others considered in fig .
[ fig : comp2 ] and bottom of table [ tab : comp1 ] .
however a word of caution must be said here , the @xmath81-wave part of the antikaon selfenergy of ref .
@xcite is based on a model for the @xmath80 scattering in the free space that , though it is quite successful near threshold , predicts amplitudes for the isoscalar channel around @xmath83 mev , with real parts which are in total disagreement ( in sign and in size , see ref .
@xcite ) with the analysis of ref . @xcite .
thus , most probably one can not trust the thph model to describe the @xmath0-dynamics at this momentum .
indeed , there is no reason either to believe more in the 2dd and iaph models , and we believe that the more reliable predictions for @xmath84 mev are those based on the ia .
we thank to e.oset and a.ramos for useful discussions .
a.j.m . acknowledges a fellowship from the university of granada . this work was supported by junta
de andaluca , dgi and feder funds ( bfm2002 - 03218 ) . | ' '' '' we calculate @xmath0-nucleus elastic differential , reaction and total cross sections for different nuclei ( @xmath1c,@xmath2ca and @xmath3pb ) at several laboratory antikaon momenta , ranging from @xmath4 mev to @xmath5 mev .
we use different antikaon - nucleus optical potentials , some of them fitted to kaonic atom data , and study the sensitivity of the cross sections to the considered antikaon - nucleus dynamics . |
in condensed matter physics , quantum phase transitions , which happen at zero temperature , have received much attention . in a quantum phase transition
the properties of a many - body system will change dramatically , when a controlling parameter changes across critical point @xcite .
the phase diagram in the one - dimensional spinless fermion model has been determinded by many researchers .
it is well known that a transition from luttinger liquid to charge - density - wave phase occurs when the repulsive interaction is twice the hopping interaction .
however , long - range coulomb interactions sometimes also play an important role in frustrated systems due to their geometrical structure .
for example , a spinless fermion , including the nearest - neighbor and next - nearest - neighbor repulsive interactions can describe the low - energy electronic state of @xmath0 double chains @xcite , and the repulsions plays a significant role in the charge - ordered - insulator - metal transition .
the phases of spinless fermion including the nearest - neighbor as well as next - nearest - neighbor repulsive interactions are more complicated than one might expect .
the model can be described by the following hamiltonian : @xmath1,\end{aligned}\ ] ] where @xmath2 is the fermionic creation ( annihilation ) operator at site @xmath3 , @xmath4 is the hopping amplitude , and @xmath5 and @xmath6 are the nearest - neighbor and next - nearest - neighbor coulomb interactions .
the fermionic hamiltonian ( [ eq1 ] ) can be mapped onto the identical hamiltonian for hardcore bosons , which is given by @xmath7,\end{aligned}\ ] ] where @xmath8 is the hardcore boson creation ( annihilation ) operator at site @xmath3 , and @xmath4 is the hopping integral and taken as the unit of energy in the paper . @xmath5 and @xmath6 are the nearest - neighbor and next - nearest - neighbor interactions .
this mapping can be done by using a jordan - wigner transformation , as both models have the same spectrum .
the phase diagram for hard - core bosons is relevant to the corresponding spinless - fermion model , as is shown in refs .
it is noted that the luttinger - liquid phase in the spinless - fermion model is relevant to the superfluid phase ( sf ) in the hard - core - boson model , and other phases have the same names in the hard - core - boson model and the spinless - fermion model .
we also focus our analysis on the model ( [ eq2 ] ) at half filling .
the hamiltonian of the case @xmath9 , can be solved exactly via bethe ansatz .
we find a transition at @xmath10 from the sf to the @xmath11 charge - density - wave ( cdw - i ) phase as @xmath5 is increased .
when @xmath12 , the phase structure becomes rich .
it is reported that there exist bond - order ( bo ) phase@xcite , which is somehow was missed in some research @xcite .
when @xmath6 is large , @xmath13 charge - density - wave phase(cdw - ii ) with a four - site unit cell appears . here , @xmath14 denotes the presence ( absence ) of a particle at a particular site
recently , due to the cross over field between quantum many - body theory and quantum - information theory , the ground - state entanglement entropy and fidelity have been used to qualify quantum phase transitions in the one - dimensional spin systems @xcite .
our goal is to combine these newly developed observables with accurate numerical calculations , and independently determine the phase diagram of the above model . in the paper , we calculate the ground - state entanglement entropy and fidelity in the half - filled hardcore boson model including the nearest - neighbor as well as the next - nearest - neighbor repulsive interactions , and use them as good tools for searching for the phase transition points .
the remainederof the paper is organized as following . the measurements and the details of the methods to obtain the ground state are introduced in sec .
the results for entanglement entropy are presented in sec .
iii and for fidelity in sec iv .
finally a discussion is given at last in sec .
the entanglement entropy can be chosen as a measurement of the bipartite entanglement , which can be used to detect a quantum phase transition .
the entropy is defined as follows .
let @xmath15 be the ground state of a chain system , which can be divided into two parts a and b. the reduced density matrix of part a can be obtained by taking the partial trace over system b , which is given by @xmath16 the bipartite entanglement between parts @xmath17 and @xmath18 can be measured via the entanglement entropy as @xmath19 another concept from quantum information theory , ground - state fidelity , can be applied to capture the existence of the quantum phase transitions .
the general hamiltonian of a quantum many - body system can be written as @xmath20 , where @xmath21 is the driving hamiltonian and @xmath22 denotes its strength .
if @xmath23 represents a state of the system , the ground - state fidelity between @xmath23 and @xmath24 can be defined as @xmath25.\ ] ] for a pure state @xmath26 , eq .
[ eq5 ] can be rewritten as @xmath27 .
@xmath28 reaches its maximum value @xmath29 at @xmath30 . by expanding the fidelity in powers of @xmath31 , and since the first derivative @xmath32 , the fidelity can be written as @xmath33 therefore , the average fidelity susceptibility @xmath34 can be given by @xcite @xmath35}{n\delta^2},\ ] ] where @xmath36 is the length of the system .
as is well known , it is not easy to calculate the ground - state fidelity and entanglement entropy , because of the lack of knowledge of the ground - state wave function .
the hamiltonian ( [ eq2 ] ) is not exactly solvable except for @xmath9 ; we resort to exact diagonalization to obtain the ground state for small size , i.e. , @xmath37 . for large size , the method of the density - matrix renormalization - group(dmrg ) @xcite can be applied to obtain the ground state of model ( [ eq2 ] ) .
the total number of density matrix eigenstates held in a system block is @xmath38 in the basis truncation procedure .
the matlab code for the finite - size density - matrix renormalization group with double precision is used to evaluate the model with system sizes up to @xmath39 , and the truncation error is smaller than @xmath40 . both the above two methods are used to calculate the model ( [ eq2 ] ) with open boundary conditions .
( color online ) the entanglement entropy of two central qubits ( a ) and its first derivative ( b ) are plotted as functions of @xmath6 for @xmath41 with @xmath42 for different system sizes . here and in the following graphics
all quantities are dimensionless .
, scaledwidth=55.0% ]
in order to avoid boundary effects , the entanglement entropy of two central two neighboring sites of a large system is calculated . by use of exact diagonalization and the dmrg ,
the entanglement entropy of the central two particles is plotted as a function of interaction for sizes @xmath43 in fig .
[ fig1](a ) , and the corresponding first derivatives are shown in fig .
[ fig1](b ) .
it is found that the entanglement entropy initially increases rapidly with increases of @xmath6 .
when the interaction @xmath6 increases further , the entanglement entropy increases slowly , and reaches a peak value at @xmath44 , which is independent of system size . as we know from refs .
@xcite , the peak of local entanglement does not coincide with the critical point .
considering the energy of the system in terms of the reduced density matrix of two spins at positions @xmath3 and @xmath45 , the energy reads @xmath46 where @xmath47 is the reduced hamiltonian of the two sites at positions @xmath3 and @xmath45 , and the sum is over the total hamiltonian of the system .
@xmath48 is the reduced density matrix of two sites at positions @xmath3 and @xmath45 .
moreover , the second - order derivative of the energy is related to the derivative of the reduced density matrix .
this is shown by @xmath49.\end{aligned}\ ] ] it is noted that a discontinuity in the second derivatives of the energy requires the divergence of at least one of the derivatives @xmath50 at the critical points @xcite.the phase transition points sometimes are determined by the positions at which the first - order derivative of the two - site entanglement entropy takes on its maximal and minimal values @xcite . here
, the cdw - ii to bo transition is a second - order transition , so it should be captured by the first - order derivative of the two - site entanglement entropy .
the phase transitions of sf to bo and bo to cdw - ii are infinite - order berezinskii - kosterlitz - thouless ( bkt ) transitions , which can also be predicted by the first - order derivative of the two - site entanglement entropy .
the first derivative of the entanglement entropy is shown in fig .
[ fig1](b ) .
there are two peaks , whose locations move to high @xmath6 with increase in system size , and the location of the valley moves to lower @xmath6 with system size increases .
these behaviors capture the cdw - i to sf , sf to bo , and bo to cdw - ii phase transitions , respectively .
( color online ) the phase diagram of the model ( [ eq2 ] ) at half filling in the tilted @xmath5-@xmath6 plane .
the dashed line corresponds to @xmath51 , which is the boundary of the cdw - i and cdw - ii phases when @xmath52 .
, scaledwidth=25.0% ] the positions of the two peaks and one valley in the first derivative of the two - site entropy can be extrapolated to the thermodynamic limit .
we investigate the size dependence of the positions of their extreme points .
the changes of the critical points can be fitted by the formula @xmath53 , where @xmath54 are constants and @xmath36 is the size of the system .
we plot the locations of the two peaks and one valley as functions of @xmath55 .
the fitting results are shown in fig .
[ fig2 ] . here
we obtain that @xmath56 , @xmath57 , @xmath58 , @xmath59 , @xmath60 , @xmath61 , and @xmath62 , @xmath63 , @xmath64 . in fig .
[ fig3 ] , we also show the phase diagram of the model , which is detected via the entanglement entropy .
it is seen that our results for the cdw - i to sf boundary and the cdw - ii to bo boundary are consistent with the results in refs .
the difference is in the bo to sf boundary .
the region of the bo phase is smaller than the region captured by the static structure factor . for @xmath65
, we capture the bo to sf critical phase transition point @xmath66 .
it is consistent with the result obtained from the energy gap and larger than the result @xmath67 obtained from the static structure factor , which is also shown in ref .
we apply the dmrg to calculate the ground - state fidelity susceptibility with system size @xmath36 up to @xmath68 and @xmath69 .
the ground - state fidelity susceptibility @xmath70 is plotted as a function of the interaction @xmath6 for different sizes in fig .
[ fig4 ] in dimensionless units .
two peaks in the fidelity susceptibility is found .
the peaks of @xmath70 increase when the system size increases .
one of the peaks locations moves to high @xmath6 up to a particular value as the system size increases , and the other moves to a slightly lower @xmath6 down to a particular value as the system size increases .
the fidelity measures the similarity between two states , while quantum phase transitions are intuitively accompanied by an abrupt change in the structure of the ground - state wave function .
this primary observation motivates researchers to use the fidelity to predict quantum phase transitions .
a peak in the fidelity susceptibility indicates a quantum phase transition . here , scaling of the extreme points of the fidelity susceptibility as the system length increases is also investigated .
we find that the changes of the maximal points can also be fitted by the formula @xmath53 , where @xmath54 are constants and @xmath36 is the number of the system .
the results for the locations of the fidelity susceptibility can be used to investigate the quantum phase transition points in the thermodynamic limit .
we plot the locations of the maximum fidelity susceptibility as a function of @xmath55 and show the numerical fit in fig .
[ fig4 ] ( b ) and [ fig4 ] ( c ) .
we obtain that @xmath71 , @xmath72 , @xmath73 , and @xmath62 , @xmath63 , @xmath64 .
the quantum phase transitions points are partially consistent with the results obtained from the entanglement entropy .
it is shown that the fidelity susceptibility can predict the cdw - i to sf transition and the bo to cdw - ii transition , however , it can not predict sf to bo transition . as we know , the bo to cdw - ii and sf to bo transitions are both infinite - order bkt transitions .
the numerical results clearly show that the bkt transition point is detectable to the bo to sf transition by divergence of the fidelity susceptibility @xcite .
our results are similar to the results , which the entanglement entropy predicts the bo phase in the extended hubbard model @xcite , and the fidelity can not @xcite .
the bkt phase transition point sometimes can not be well characterized by the ground - state fidelity for finite - size systems @xcite .
a other reason may be that , according to the analysis of ref .
@xcite , a divergence in the fidelity susceptibility implies a quantum phase transition , but the converse is not true .
this means that there are quantum phase transitions driven by particularly weak perturbations , where the fidelity susceptibility may not diverge .
it would be interesting to use infinite - time - evolving block decimation@xcite to calculate the ground - state fidelity , and use it to detect the superfluid to bond order transition .
here we do not show the phase diagram of the model which can be detected by using the fidelity susceptibility .
( color online ) ( a ) fidelity susceptibility versus interaction @xmath6 for @xmath41 with @xmath42 for different system sizes .
( b),(c ) the scaling behavior of the extreme points of @xmath74 versus @xmath55 .
the red lines are fixed lines.,scaledwidth=50.0% ]
in the paper , we employ the entanglement entropy to study numerically the hard - core - boson model with nearest and next - nearest - neighbor interactions . by using exact diagonalization and the density - matrix renormalization group ,
the effects of interactions on the entanglement entropy in the model are investigated .
it is found that the first derivative of the entanglement entropy can detect all the quantum phase transitions in the model , and the phase diagram is given .
we also test the ability of the fidelity susceptibility to detect the phase transitions .
it is found that the fidelity susceptibility can predict charge - density - wave ii to bond order and charge - density - wave i to superfluid transitions , however , the fidelity susceptibility can not predict the superfluid to bond order transition for finite - size systems .
j. r. thanks wen - long you and yin - zhong wu for their helpful discussions . financial supports from the national natural science foundation of china ( grants
no:11104021 , 11174114 , and 11274054 ) is gratefully acknowledged . | the entanglement entropy and quantum fidelity in a hard - core - boson model with nearest- and next - nearest - neighbor interactions are studied numerically . by using exact diagonalization and the density matrix renormalization group ,
the effects of interactions on entanglement entropy and fidelity susceptibility in the model are investigated .
we focus our attention on looking for three quantum phase transitions .
it is found that the first derivative of the entanglement entropy can indicate all of three phase transitions , while the fidelity susceptibility can not predict the transition from superfluid to bond - order in a finite - size system . |
the quest for @xmath0 , the current ratio of the mean mass density of the universe to the critical density required for closure , has been a focus of the research efforts of many astrophysicists involving a variety of different techniques . at present , most observational evidence suggests a universe with sub - critical matter density , perhaps with a cosmological constant making up the difference required for a critical universe ( e.g. , coles & ellis 1994 ; ostriker & steinhardt 1995 ) .
the possibility of measuring @xmath0 using the amount of `` substructure '' in galaxy clusters has thus generated some interest , `` this is a critical area for further research , as it directly tests for @xmath15 in dense lumps , so both observational and theoretical studies on a careful quantitative level would be well rewarded . ''
( ostriker 1993 ) .
early analytical work ( e.g. , richstone , loeb , & turner 1992 ) and simulations ( evrard et al . 1993 ; mohr et al . 1995 ) found that the morphologies of x - ray clusters strongly favored @xmath16 over low - density universes .
along with @xmath17 analysis of cosmic velocity fields ( e.g. , dekel 1994 ) , these substructure analyses were the only indicators in support of a critical value of @xmath0 . however , the analytical results ( e.g. , kauffmann & white 1993 ; nakamura , hattori , & mineshige 1995 ) , simulations ( e.g. , jing et al .
1994 ) , and morphological statistics ( e.g. , buote & tsai 1995b ) have been criticized rendering the previous conclusions about @xmath0 uncertain .
buote & tsai ( 1995b , hereafter bta ) introduced the power ratios ( prs ) for quantifying the spatial morphologies of clusters in terms of their dynamical states .
the prs essentially measure the square of the ratio of a higher order moment of the two - dimensional gravitational potential to the monopole term computed within a circular aperture , where the radius is specified by a metric scale ( e.g. , 1 mpc ) .
buote & tsai ( 1996 , hereafter btb ) computed prs of @xmath11 x - ray images for a sample of 59 clusters and discovered that the clusters are strongly correlated in pr space , obeying an `` evolutionary track '' which describes the dynamical evolution of the clusters ( in projection ) .
tsai & buote ( 1996 , hereafter tb ) studied the prs of a small sample of clusters formed in the hydrodynamical simulation of navarro , frenk , & white ( 1995 ) and verified the interpretation of the `` evolutionary track '' .
in contrast to the previous studies ( e.g. , richstone et al .
1992 ; mohr et al .
1995 ) , tb concluded that their small cluster sample , formed in a standard @xmath14 , cdm simulation , possessed too much substructure ( as quantified by the prs ) with respect to the @xmath11 clusters , and thus favored a lower value of @xmath0 .
however , a statistically large sample of clusters is important for studies of cluster morphologies .
the prs are most effective at categorizing clusters into different broad morphological types ; i.e. the distinction between equal - sized bimodals and single - component clusters is more easily quantified than are small deviations in ellipticities and core radii between single - component clusters ( see bta ) .
the efficiency of the prs at classifying clusters into a broad range of morphological types is illustrated by their success at quantitatively discriminating the @xmath11 clusters along the lines of the morphological classes of jones & forman ( 1992 ) ( see btb ) .
there is a lower frequency of nearly equal - sized bimodals in the @xmath11 sample than clusters with more regular morphologies .
hence , to make most effective use of the prs the models need to be adequately sampled ( i.e. simulations have enough clusters ) to ensure that relatively rare regions of pr - space are sufficiently populated . in this paper we build on the previous studies and investigate the ability of the prs to distinguish between models having different values of @xmath0 .
unlike the previous theoretical studies of cluster morphologies mentioned above , we also consider models having different power spectra , @xmath1 , since @xmath1 should affect the structures of clusters as well . at the time we began this project it was too computationally costly to use hydrodynamical simulations to generate for several cosmological models a large , statistically robust , number of clusters with sufficient resolution . to satisfy the above criteria and computational feasibility we instead used pure n - body simulations .
the organization of the paper is as follows .
we discuss the selection of cosmological models in [ models ] ; the specifications of the n - body simulations in [ sample ] ; the validity of using dark - matter - only simulations to generate x - ray images and the construction of the images in [ xray ] ; and computation of the prs in [ prs ] .
we analyze the models having different values of @xmath0 and a cosmological constant in
[ omega ] , and models with different spectral slopes and @xmath5 in [ spectrum ] .
the implications of the results for all of the models and comparison of the simulations to the @xmath11 sample of btb is discussed in
[ disc ] . finally , in [ conc ] we present our conclusions .
to test the sensitivity of cluster morphologies to the cosmological density parameter due to matter , @xmath0 , and the power spectrum of density fluctuations , @xmath1 , we examined several variants of the standard cold dark matter ( cdm ) model ( e.g. , ostriker 1993 ) . in table
[ table.models ] we list the models and their relevant parameters : @xmath0 ; @xmath18 , where @xmath19 is a cosmological constant and @xmath20 is the present value of the hubble parameter ; the spectral index , @xmath6 , of the scale - free power spectrum of density fluctuations , @xmath7 ; and @xmath5 , the present rms density fluctuations in spheres of radius @xmath21 mpc , where @xmath22 is defined by @xmath23 km s@xmath24 mpc@xmath24 .
the parameters of the open cdm model ( ocdm ) and low - density , flat model ( lcdm ) were chosen to be consistent with current observations ( e.g. , ostriker & steinhardt 1995 ) . their normalizations were set according to the @xmath25 relationship of eke , cole , & frenk ( 1996 ) to agree with the observed abundance of x - ray clusters .
the biased cdm model ( bcdm ) was also normalized in this way .
however , the bcdm simulation , because it has @xmath14 , necessarily has poorer resolution ( i.e. fewer particles per cluster ) than the ocdm and lcdm models due to the fixed box size of our simulations ( see [ xray ] ) . for the purposes of our investigation of cluster morphologies it is paramount to compare simulations having similar resolution .
hence we use the scdm model ( with @xmath26 ) as our primary @xmath14 simulation for analysis , which has resolution equivalent to the ocdm and lcdm simulations .
( we show in [ svsb ] that the means of the pr distributions for bcdm and scdm are very similar which turns out to be most important for examining the effects of @xmath0 . ) hence , the scdm , ocdm , and lcdm models allow us to explore the effects of @xmath0 and @xmath27 on the cluster morphologies ; comparing scdm and bcdm provides information on the influence of @xmath5 .
we explore the effects of different @xmath1 on the prs using the scale - free models , which have different @xmath6 from scdm .
for the scale - free models we normalized each to the same characteristic mass , @xmath28 , defined to be ( cole & lacey 1996 ) the mass scale when the linear rms density fluctuation is equal to @xmath29 , the critical density for a uniform spherically symmetric perturbation to collapse to a singularity . for @xmath14 ,
the linear theory predicts @xmath30 ( e.g. , padmanabhan 1993 ) .
we take the scdm model with @xmath26 as a reference for these scale - free models which gives a characteristic mass of @xmath31 .
this procedure allows a consistent means to normalize the scale - free models relative to each other on the mass scales of clusters .
unfortunately , as a result of this normalization procedure , at earlier times the models have different large - scale power and thus the cluster mass functions are different for each of the models .
the scale - free model with @xmath32 is similar to the scdm model and will be used to `` calibrate '' the scale - free models with respect to the other models ( see table [ table.models ] ) .
we use the tree - particle - mesh ( tpm ) n - body code ( xu 1995b ) to simulate the dissipationless formation of structure in a universe filled with cold dark matter .
the simulations consist of @xmath33 particles in a square box of width @xmath34 mpc .
the gravitational softening length is @xmath35 kpc which translates to a nominal resolution of @xmath36 kpc .
this resolution is sufficient for exploring the structure of clusters with prs in apertures of radii @xmath37 mpc ; for a discussion of the related effects of resolution on the performance of prs on @xmath11 x - ray images see buote & tsai ( 1995b , 4 ) .
all of the realizations have the same initial random phase . for each simulation we located the 39 most massive clusters using a version of the denmax algorithm ( bertschinger & gelb 1991 ) modified by xu ( 1995a )
. this convenient selection criterion yields well defined samples for each simulation and allows consistent statistical comparison between different simulations which is the principal goal of our present investigation . for the various cosmological models we explore ( see table [ table.models ] ) these clusters generally have masses ranging from @xmath38 , which correspond to typical cluster masses observed in x - ray ( e.g. , edge et al .
1990 ; david et .
al . 1993 ) and optically ( e.g. , carlberg et .
al . 1995 ) selected samples . by letting the gas density trace the dark matter density @xmath40 and by assuming that the plasma emissivity of the gas is constant , we computed the x - ray emissivity of the clusters , @xmath39 .
given its importance on the results presented in this paper , here we discuss at some length the suitability of this approximation .
( cooling of the gas is discussed in [ rosat ] . ) for clusters in the process of formation or merging , the gas can have hot spots appearing where gas is being shock heated ( e.g. , frenk , evrard , summers , & white 1996 ) .
one effect of such temperature fluctuations on the intrinsic x - ray emissivity is that the intrinsic plasma emissivity will vary substantially over the cluster thus rendering @xmath39 a poor approximation .
however , the intrinsic x - ray emissivity is not observed , but rather that which is convolved with the spectral response of the detector .
for @xmath11 observations of clusters with the pspc , the plasma emissivity is nearly constant over the relevant ranges of temperatures ( nra 91-ossa-3 , appendix f , @xmath11 mission description ) , and thus temperature fluctuations contribute negligibly to variations in the emissivity ; for previous discussions of this issue for the prs see bta and tb .
a more serious issue is whether the shocking gas invalidates the @xmath41 approximation , in which case the dynamical state inferred from the gas would not reflect that of the underlying mass .
tb , who analyzed the hydrodynamical simulation of navarro et al .
( 1995a ) , showed that the prs computed for both the gas and the dark matter gave similar indications of the dynamical states of the simulated clusters ( see 4 of tb ) .
in particular , this applied at early times when the clusters underwent mergers with massive subclusters .
hence , our approximation for the x - ray emissivity should be reasonable even during the early , formative stages of clusters .
another possible concern with setting @xmath41 is that a gas in hydrostatic equilibrium , which should be a more appropriate description for clusters in the later stages of their evolution , traces the shape of the potential of the gravitating matter which is necessarily rounder than the underlying mass ; if the gas is rounder , then the prs will be _
smaller_. however , the core radius ( or scale length ) of the radial profile of the gas also influences the prs .
in fact , clusters with larger core radii have _ larger _ prs ; see bta who computed prs for toy x - ray cluster models having a variety of ellipticities and core radii .
when isothermal gas , which is a good approximation for a nearly relaxed cluster , is added to the potential generated by an average cluster formed in a @xmath14 , cdm simulation , the gas necessarily has a _ larger _ core radius than that of the dark matter ( see figure 14 of navarro , frenk , & white 1995b ) .
hence , a gas in hydrostatic equilibrium will have a larger core radius than that of the dark matter , at least in the context of the cdm models we are studying . considering the competing effects of smaller ellipticity and larger core radii ( factors of 2 - 3 in each ) , and from consulting table 6 of bta , we conclude that no clear bias in the prs is to be expected by assuming that the gas follows the dark matter . in further support of this conclusion
are the similarities of the morphologies of clusters in the n - body study of jing et al .
using centroid - shifts and axial ratios they find similar results when @xmath41 and when the gas is in hydrostatic equilibrium ( see their figures 5 , 6 , and 8)
. having chosen our representation for the x - ray emissivity , we then generated two - dimensional `` images '' for each cluster .
a rectangular box of dimensions @xmath42 @xmath43 mpc@xmath44 with random orientation was constructed about each cluster .
we converted the particle distribution for each cluster to a mass density field using the interpolation technique employed in smoothed particle hydrodynamics ( sph ) ( e.g. , hernquist & katz 1989 ) , from which the x - ray emissivity was generated , @xmath39 .
the sph interpolation calculates the density at a grid point by searching for the nearest neighbors and is thus more robust and physical than other linear interpolation schemes like cloud - in - cell . for our sph interpolation
we use 20 neighbors and the spline kernel described in hernquist & katz .
the interpolation result is independent of the cell we choose for the x - ray emissivity calculations .
typically , the boxes contained @xmath45 particles for a cluster ; e.g. , scdm ( 1799 - 3965 ) , ocdm ( 1359 - 3656 ) , lcdm ( 1439 - 3877 ) , and bcdm ( 603 - 1740 ) .
we projected the emissivity along the long edge of the box into a square @xmath46 @xmath47 mpc@xmath48 `` image '' consisting of `` pixels '' of width @xmath49 kpc .
this pixel width was chosen sufficiently small so as not to inhibit reliable computation the prs .
we do not add statistical noise or other effects associated with real observations to the x - ray images since our principal objective is to examine the intrinsic response of cluster morphologies to different cosmological parameters .
however , the investigation of observational effects on the prs by bta , and the derived error bars on the prs from @xmath11 clusters by btb , do not show any large systematic biases ; comparison of the simulations to the @xmath11 cluster sample is discussed in [ disc ] .
the prs are derived from the multipole expansion of the two - dimensional gravitational potential , @xmath50 , generated by the mass density , @xmath51 , interior to @xmath52 , @xmath53 where @xmath54 is the azimuthal angle , @xmath55 is the gravitational constant and , @xmath56 because of various advantageous properties of x - ray images of clusters , we associate the surface mass density , @xmath57 , with x - ray surface brightness , @xmath58 ( which is derived from the projection of @xmath13 see previous section ) ; for more complete discussions of this association see bta and tb . the square of each term on the right hand side of eq .
( [ eqn.multipole ] ) integrated over the boundary of a circular aperture of radius @xmath59 is given by ( ignoring factors of @xmath60 ) , @xmath61 for @xmath62 and , @xmath63 ^ 2,\label{eqn.power0}\ ] ] for @xmath64 .
it is more useful for studies of cluster structure to consider the ratios of the higher order terms to the monopole term , @xmath65 , which we call `` power ratios '' ( prs ) . by dividing each term by @xmath66 we normalize to the flux within @xmath59 . because for clusters , @xmath67 for @xmath62 ( e.g. , btb ) , it is preferable to take the logarithm of the prs , @xmath68 which we shall henceforward analyze in this paper .
since the @xmath69 depend on the origin of the chosen coordinate system , we consider two choices for the origin .
first , we take the aperture to lie at the centroid of @xmath58 ; i.e. where @xmath70 vanishes .
of these centroided @xmath71 , @xmath72 , @xmath73 , and @xmath74 prove to be the most useful for studying cluster morphologies ( see bta ) . in order to extract information from the dipole term
, we also consider the origin located at the peak of @xmath58 .
we denote this dipole ratio by @xmath75 , and its logarithm @xmath76 , to distinguish it from the centroided power ratios . to obtain the centroid of @xmath58 in a consistent manner for all clusters we adopted the following procedure .
first , when projecting the cluster ( see [ xray ] ) , the cluster was roughly centered on the x - ray image by eye . for each image we computed the centroid in a circular aperture with @xmath77 @xmath78 mpc located about the field center .
this centroid was then used as our initial center for each cluster ; see bta for a description of how the peak of @xmath58 is located .
in addition to considering the @xmath71 individually , we also analyze the cluster distributions along the `` evolutionary tracks '' in the @xmath79 and @xmath80 planes obeyed by the @xmath11 clusters of btb .
we refer the reader to tb for a detailed discussion of the cluster properties along the evolutionary tracks . using the augmented edge et al .
( 1990 ) sample of btb we recomputed the lines defining the evolutionary tracks of the @xmath11 data for the @xmath81 mpc apertures .
this was done since tb selected a subset of the clusters based on @xmath71 measurement uncertainty rather than flux .
following tb we fit @xmath82 considering the uncertainties in both axes ; a similar fit was done for the @xmath80 plane .
we obtained @xmath83 , @xmath84 for the @xmath79 track which we denote by @xmath85 .
similarly , for the @xmath80 track , which we denote by @xmath86 , we obtained @xmath87 , @xmath88 .
these results are nearly the same found by tb for the slightly different sample .
to facilitate comparison to a previous study of the @xmath71 of @xmath11 clusters ( btb ) , we compute @xmath71 of the simulated clusters in apertures ranging in radius from @xmath89 mpc to @xmath90 mpc ( @xmath91 km s@xmath24 mpc@xmath24 ) in steps of @xmath92 mpc ; i.e. @xmath93 mpc in steps of @xmath94 mpc .
we refer the reader to 2 of bta and 2 of tb for discussions of the advantages of using a series of fixed metric aperture sizes to study cluster morphologies .
first we consider clusters formed in the scdm , ocdm , and lcdm models .
contour plots for 16 of the clusters formed in each of the models are displayed in figures [ fig.scdm ] , [ fig.ocdm ] , and [ fig.lcdm ] . in figure [ fig.p2p4cdm ]
we show the @xmath95 plane for the @xmath89 mpc and @xmath96 mpc apertures ; the scdm model appears in each plot for comparison .
the clusters in each of the models exhibit tight correlations very similar to the evolutionary tracks of the @xmath11 clusters ( btb ) and the simulated hydrodynamic clusters @xmath2 of navarro et al .
( 1995a ) studied by tb . along the evolutionary tracks a shift in the means of the @xmath71 is easily noticeable in the @xmath89 mpc aperture , being most apparent for the scdm - ocdm models .
the spread of the @xmath71 along the track in the @xmath96 mpc aperture for scdm - ocdm also appears to be different .
the distributions perpendicular to @xmath85 do not show discrepancies obvious to the eye . at this time we shift our focus away from the evolutionary tracks and instead analyze the individual @xmath71 distributions , which prove to be more powerful for distinguishing between cosmological models as we show below .
we give the individual @xmath71 distributions of clusters in the three models for the @xmath89 mpc and @xmath96 mpc apertures in figures [ fig.prhist.05mpc ] and [ fig.prhist.10mpc ] .
we found it most useful to compare these distributions in terms of their means , variances , and kolmogorov - smirnov ( ks ) statistics .
for the number of clusters in each of our simulations ( 39 ) higher order statistics like the skewness and kurtosis are unreliable `` high variance '' distribution shape estimates ( e.g. , bird & beers 1993 ) .
we did consider more robust statistics like the `` asymmetry index '' ( ai ) , which measures a quantity similar to the skewness , and the `` tail index '' ( ti ) , which is similar to the kurtosis ( bird & beers 1993 ) .
however , we found that they did not clearly provide useful information in addition to the lower order statistics and ks test , and thus we do not discuss them further .
the means and standard deviations for the @xmath97 mpc apertures are listed in table [ table.avgpr ] . as a possible aid to understanding the relationships between the values in table [ fig.avgpr.omega ] ,
we plot in figure [ fig.avgpr.omega ] the standard deviation vs the mean for the @xmath98 mpc aperture .
we do not present the results for the larger apertures because they did not significantly improve the ability to distinguish between the models .
moreover , we found that @xmath74 and @xmath76 do not provide much useful information in addition to @xmath72 and @xmath73 .
generally @xmath74 tracks the behavior of @xmath72 , though showing less power to discriminate between models ; the similarity to @xmath72 is understandable given the strong correlation shown in figure [ fig.p2p4cdm ] .
likewise , @xmath76 is similar to , but not quite so effective as , @xmath73 . for compactness , thus , we shall henceforward mostly restrict our discussion to results for @xmath72 and @xmath73 in the @xmath97 mpc apertures .
we compare the means , standard deviations , and total distributions of the models in table [ table.test ] using standard non - parametric tests as described in press et al .
( 1994 ) . the student s t - test , which compares the means of two distributions , computes a value , @xmath99 , indicating the probability that the distributions have significantly different means .
similarly , the f - test , which compares the variances of two distributions , computes a value , @xmath100 , indicating the probability that the distributions have significantly different variances .
finally , the ks test , which compares the overall shape of two distributions , computes a value , @xmath101 , indicating the probability that the distributions originate from the same parent population ; the probabilities listed in table [ table.test ] are given as percents ; i.e. decimal probability times 100 .
note that for the cases where the f - test gives a probability less than 5% we use the variant of the t - test appropriate for distributions with significantly different variances ( i.e. program _
tutest _ in press et .
al . ) . as is clear from inspection of figures [ fig.p2p4cdm ] - [ fig.avgpr.omega ] , and tables [ table.avgpr ] - [ table.test ] ,
the means of the @xmath71 of the scdm model exceed those of ocdm . in terms of the t - test
the significance of the differences is very high .
of all the @xmath71 , generally the means of @xmath72 and @xmath73 exhibit the largest significant differences ; the most significant differences are seen for @xmath73 in the @xmath89 mpc aperture , @xmath102 , and for @xmath72 in the @xmath98 mpc aperture , @xmath103 .
hence , though different in all the apertures , the discrepancy in the means is most significant for the smallest apertures , @xmath104 mpc
. the variances of @xmath73 in the scdm model are essentially consistent at all radii with their corresponding values in ocdm .
however , for @xmath72 the variances are consistent at @xmath98 mpc , but marginally inconsistent at @xmath105 mpc ( and inconsistent at @xmath106 mpc ) .
the ks test generally indicates a significant difference in scdm and ocdm when also indicated by the t - test , or the t - test and f - test together .
the level of discrepancy is usually not as significant as given by the t - test , except when @xmath100 is small as well .
since the ks test does not indicate discrepancy when both the t - test and f - test indicate similarity , we conclude that higher order properties of the pr distributions are probably not very important for the scdm and ocdm models ( at least for the samples of 39 clusters in our simulations ) . since this qualitative behavior holds for the other model comparisons , we shall not emphasize the ks tests henceforward . finally , in terms of the various significance tests we find that the @xmath85 distribution essentially gives a weighted probability of the individual @xmath72 and @xmath74 distributions ; i.e. it does not enhance the discrepancy in the individual distributions .
perpendicular to @xmath85 the distributions are consistent . the same behavior is seen for @xmath86 as well .
this behavior is seen for the remaining model comparisons in this section so we will not discuss the joint distributions further .
the @xmath71 means for the scdm clusters also systematically exceed those in the lcdm model , however the significance of the difference is not as large as with the ocdm clusters .
the largest discrepancy is observed for @xmath72 in the @xmath107 mpc apertures for which @xmath108 .
the other @xmath71 show only a marginal discrepancy in the means . for apertures
@xmath104 mpc , @xmath73 has @xmath109 , but is quite consistent at larger radii . the variances for the scdm and lcdm models are consistent for essentially all radii and all @xmath71 .
the @xmath71 means for the lcdm clusters appear to systematically exceed those in the ocdm model , however the formal significances of the differences are quite low .
the means are entirely consistent at all radii for @xmath72
. however , @xmath73 shows a marginal difference in the @xmath89 mpc aperture ( @xmath110 ) .
the variances of the @xmath71 of the ocdm and lcdm models behave similarly as with the scdm and ocdm comparison above , as expected since the scdm - lcdm variances are essentially identical . however , the degree of discrepancy is not as pronounced .
the means of the individual @xmath71 distributions generally exhibit the most significant differences between the scdm , ocdm , and lcdm models ; the variances are much less sensitive to the models , with @xmath73 showing no significant variance differences
. the larger means for the @xmath71 in the scdm models are expected from the arguments of , e.g. , richstone et al .
that is , in a sub - critical universe the growth of density fluctuations ceased at an early epoch and so present - day clusters should show less `` substructure '' than in an @xmath14 universe where formation continues to the present .
clusters with more structure will have systematically larger values of the @xmath71 .
the @xmath71 whose means show the most significant differences between the models are @xmath72 and @xmath73 , where @xmath72 typically performs best for apertures @xmath107 mpc and @xmath73 is most effective for @xmath104 mpc . although
useful , @xmath76 is often the least effective @xmath71 for differentiating models in terms of its mean ; this relatively weak performance of the dipole ratio with respect to other moments is echoed in the results of jing et al .
( 1995 ) who found that their measure of an axial ratio performed better than a centroid shift for discriminating between models ( see their tables 3 - 6 ) .
in this section we investigate the effects of different power spectra for models otherwise conforming to the specifications of the scdm model .
first , we examine models with different spectral indices of the scale - free power spectrum @xmath111 , @xmath112 .
then we examine the bcdm model which has a lower power - spectrum normalization as expressed by @xmath5 . as in the previous section ,
we find the @xmath97 mpc apertures to be more useful than the larger apertures , and that @xmath74 and @xmath76 do not provide much useful information in addition to that provided by @xmath72 and @xmath73 .
hence , for compactness we again mostly restrict the discussion to @xmath72 and @xmath73 in the smaller apertures . before analyzing the @xmath71 of models with different @xmath6
we calibrate the scale - free models by comparing the @xmath32 scale - free model to the scdm model since they should have similar properties ( see [ models ] ) .
we find that the means , variances , and ks statistics of the centroided @xmath71 for the scdm and @xmath32 models are entirely consistent for all aperture radii with only one possible exception .
the variances of @xmath73 exhibit a marginal @xmath113 discrepancy in the @xmath89 mpc aperture .
the significance of this variance discrepancy should be treated with caution given the complete consistency of the means @xmath114 and ks ( 32% ) test at this radius as well as the consistency of all the tests at all the other radii investigated .
hence , the cluster morphologies of the scdm and @xmath32 models are very consistent expressed in terms of the centroided @xmath71 ( @xmath115 ) . in figure [ fig.ets.sf ]
we plot for the @xmath118 models the pr correlations for @xmath119 and @xmath120 in the @xmath105 mpc apertures .
histograms for the individual @xmath71 in these apertures are displayed in figures [ fig.sfhist.05mpc ] and [ fig.sfhist.10mpc ] .
table [ table.avgpr ] lists the means , figure [ fig.avgpr.pk ] plots the standard deviations versus the averages of the @xmath71 in the @xmath98 mpc aperture , and table [ table.test ] gives the results of the significance tests .
the means of @xmath73 are very consistent for the @xmath118 models at all radii examined .
those for @xmath72 may show some differences in their means , with the @xmath117 models perhaps having systematically smaller values .
the significance of the different means for @xmath72 is only formally marginal , with @xmath121 for aperture radii @xmath97 mpc .
however , the possible small differences in the means of @xmath72 are dwarfed by the corresponding highly significant differences in its variances .
generally the variances for all the @xmath71 in all the apertures are smaller for the @xmath117 clusters .
the most significant variance differences are observed for @xmath72 which has @xmath122 in @xmath107 mpc apertures and @xmath123 for @xmath89 mpc .
the variances for @xmath73 show differences but at a lower level of significance and only in the @xmath97 mpc apertures ; i.e. @xmath124 .
similar to what we found in
[ omega ] , the differences implied by the ks test generally follow the significances implied by the t - test and f - test ; i.e. higher order effects in the distributions are probably not overly important ( at least for our sample sizes of 39 clusters ) .
moreover , again we find that analysis of the @xmath71 in terms of the evolutionary tracks does not add useful information to the previous results .
the mean and variance effects for the individual @xmath71 translate to very similar behavior along @xmath85 .
the direction perpendicular to @xmath85 is essentially consistent for all of the tests . as a result
, we do not emphasize the ks tests or the evolutionary tracks further .
the behavior for other @xmath6 is similar , but depends to some extent on the range examined .
we find that the range of @xmath6 which accentuates differences in the @xmath71 is between @xmath125 .
over the range @xmath125 the discrepancy of means for @xmath72 essentially follows that of the full @xmath118 discussed in [ n0n2 ] .
however , the variances are not so highly discrepant as before , with @xmath126 for @xmath72 for aperture radii @xmath104 mpc ; elsewhere the variances of @xmath72 are consistent between the @xmath125 models . over the @xmath125 range
@xmath73 is consistent for all statistics at all radii examined .
the @xmath71 exhibit very few differences over the range of indices @xmath127 . for all radii examined the means and ks statistics are consistent for all the @xmath71 .
however , the variances do show some marginal differences .
the @xmath96 mpc apertures has the most significance difference where @xmath128 for both @xmath72 and @xmath73 .
also in the @xmath98 mpc aperture @xmath72 has @xmath129 .
otherwise the variances of these @xmath71 are consistent .
( we mention that the variance of @xmath72 for the @xmath32 model in the @xmath98 mpc aperture lies above that for the @xmath130 model in figure [ fig.avgpr.pk ] , but the difference is not statistically significant . )
now we consider the @xmath14 , cdm model with a lower normalization , @xmath131 , which we refer to as the biased cdm model , bcdm .
the means and variances for the bcdm model are listed in table [ table.avgpr ] , the standard deviation versus the average @xmath71 in the @xmath98 mpc aperture are plotted in figure [ fig.avgpr.pk ] , and the results for the significance tests in comparison to scdm are given in table [ table.test ] .
the means of all the @xmath71 at all aperture radii are consistent for the scdm and bcdm models .
the @xmath71 variances of the scdm clusters generally exceed those of the bcdm clusters .
the significance levels of the differences are only marginal @xmath132 and appear to be most important in the @xmath98 mpc aperture .
it is possible that the slight variance differences between the scdm and bcdm models are due to the difference in resolution between the two simulations ; i.e. the clusters in the bcdm simulations contain about half the number of particles of the scdm clusters .
we would expect that the effects of resolution would be most important in the smallest apertures ( which we do observe ) , although we would probably expect that the means as well as the variances would be affected ( which we do not observe ) .
we mention that the bcdm model performs virtually identically to the scdm model when compared to the ocdm and lcdm models .
the variances of the @xmath71 show the most significant differences between models with different power spectra ; @xmath72 generally has the most sensitive variances over the parameter ranges explored .
decreasing @xmath6 and @xmath5 both decrease the @xmath71 variances , the differences being of similar magnitude for the @xmath125 models and the scdm and bcdm models .
the means of the @xmath71 are much less sensitive to the models with different @xmath6 and @xmath5 , with @xmath72 showing the largest significant differences which are always less than differences in the variances .
no significant differences in the means are observed for @xmath73 over the range of power spectra studied .
the predominant effect of the power spectrum on the variances of the @xmath71 is intriguing .
it is reasonable that when the amount of small - scale structures is reduced ( smaller @xmath6 ) or the population of cluster - sized structures is made more uniform ( smaller @xmath5 ) that the @xmath71 distributions would also be more uniform .
the observed low sensitivity of the @xmath71 means to the power spectra is also reasonable since on average the @xmath71 means should only be affected by the rate of mass accretion through the aperture of radius @xmath59 , not by the sizes of the individual accreting clumps .
in the previous sections we have seen that differences in @xmath0 and @xmath1 in cdm models are reflected in the spatial morphologies of clusters when expressed in terms of the @xmath71 . for the purposes of probing @xmath0 , our analysis indicates that @xmath73 is the best pr since its mean is quite sensitive to @xmath0 but very insensitive to @xmath1 .
it is advantageous to also consider @xmath72 when a cosmological constant is introduced since its means differ for the scdm and lcdm models by @xmath133 whereas @xmath73 only distinguishes the models at the @xmath134 level .
the marginal dependence of the mean of @xmath72 on @xmath1 is not overly serious for studying differences in @xmath0 because the differences in means due to @xmath1 are always accompanied by larger , more significant differences in the variances ; i.e. different means for @xmath72 but consistent variances should reflect differences only in @xmath0 .
the best apertures for segregating models are generally @xmath97 mpc .
a few previous studies have examined the influence of @xmath0 and @xmath27 on the morphologies of galaxy clusters .
perhaps the most thorough investigation is that of jing et .
1995 ) who used n - body simulations of a variety of cdm models , including versions similar to our scdm , ocdm , and lcdm , to study variations of center - shifts and axial ratios .
al . reached the same qualitative conclusions as we do ; i.e. the scdm model is easily distinguished from ocdm and lcdm because it produces clusters with much more irregular morphologies than than the others .
however , jing et al .
obtained infinitesimal ks probabilities for the axial ratio when comparing scdm to ocdm and lcdm , a level of significance orders of magnitude different from that found in this paper .
the source of this discrepancy is unclear given the qualitative similarities of their axial ratio and our @xmath72 .
the disagreement may arise from differences in numerical modeling between the simulations ; i.e. the results of jing et al . are derived from simulations with a larger force resolution ( @xmath135 mpc ) , and smaller particle number for the non - scdm models ( @xmath136 ) than in our simulations , and have clusters which visually do not show the rich structures seen in our simulations ( figures [ fig.scdm ] , [ fig.ocdm ] , and [ fig.lcdm ] ) .
the qualitative results of jing et al .
agree with the hydrodynamic simulations of mohr et al .
( 1995 ) who also used center shifts and axial ratios as diagnostics for `` substructure '' . if we visually estimate the means of the center shifts and axial ratios from figures 6 and 8 of jing et al . for their scdm , ocdm , and lcdm models ( actually ocdm with @xmath137 and lcdm with @xmath138 )
, we find that they agree quite well with the corresponding values in table 3 of mohr et al . ; i.e. the results from the n - body and hydrodynamic simulations are very similar , despite the many other differences between the simulations ( e.g. , large number of baryons in ocdm clusters for mohr et al . ) .
we can make a similar comparison of the @xmath71 derived in this paper with the results from tb who analyzed the small sample of scdm clusters formed in the hydrodynamic simulation of navarro et al .
( 1995a ) .
we find that the means ( and variances ) of the @xmath71 computed in this paper are very similar to those of the hydrodynamic clusters ; e.g. , the mean for @xmath85 for @xmath81 mpc may be read off figure 7 of tb which shows excellent agreement with the scdm value we obtain from the n - body simulations ( average @xmath139 ) .
the quantitative similarity between the results , particularly between the means of the morphological statistics , for the n - body and hydrodynamic simulations of ( jing et al.,mohr et al . ) and ( this paper , tb ) suggest that it is useful to compare the @xmath71 derived from n - body simulations directly to the x - ray data .
among the biases that need to be considered in such a comparison are the effects of cooling flows ( e.g. , fabian 1994 ) , selection , and noise . cooling flows increase the x - ray emission in the cluster center , which has the effect on the @xmath71 of essentially decreasing the core size of the cluster .
judging by the observed core radii of `` regular '' x - ray clusters we would expect at most a factor of @xmath140 difference in core radii ( e.g. , a401 vs. a2029 in buote & canizares 1996 ; also see jones & forman 1984 ) .
changing the core radius by a factor of 2 typically changes @xmath72 ( for example ) by a small fraction of a decade ( see table 6 of bta ) ; this behavior , as we show below , is confirmed using a more thorough treatment .
the issue of biases between x - ray - selected and mass - selected samples needs to be addressed with hydrodynamical simulations .
the estimated uncertainties of the @xmath71 for the @xmath11 cluster sample of btb , which take into account noise and unresolved sources , do not show any clear biases .
in figure [ fig.ets.ros ] we display the correlations of the centroided @xmath71 for the @xmath11 sample of btb in the @xmath105 mpc apertures ; the scdm clusters are also plotted for a comparison .
@xmath71 histograms for these apertures are shown in figure [ fig.roshist.05mpc ] and [ fig.roshist.10mpc ] , along with those for the scdm and ocdm models .
we list the means and variances for the @xmath11 clusters in table [ table.ros.avg ] ; we plot in figure [ fig.avgpr.rosat ] the standard deviations versus the means for the @xmath11 clusters and models in the @xmath89 mpc aperture ; the results of the significance tests between the @xmath11 clusters and model clusters are given in table [ table.ros.test ] .
we analyze the @xmath11 clusters corresponding to the `` updated edge et .
1990 ) '' flux - limited sample in btb which gives 37 and 27 clusters respectively for the @xmath105 mpc apertures ; note that all the qualitative features of the results we obtain below are reproduced when all of the clusters studied in btb are used ( i.e. 59 and 44 clusters respectively ) .
the means of the scdm clusters exceed those of the @xmath11 sample to a high level of significance , with the differences being most pronounced in the @xmath89 mpc aperture .
the most significant discrepancy is for @xmath73 in the @xmath89 mpc aperture for which @xmath141 .
the variances for all the @xmath71 except @xmath73 are also significantly different , with the variances of the scdm clusters exceeding those of the @xmath11 clusters .
the scdm model has @xmath26 which is too high to fit other observations ( e.g. , ostriker & steinhardt 1995 ) .
the bcdm model , which has @xmath131 , does have @xmath71 variances in better agreement with the @xmath11 sample .
however , the means are in essentially the same level of disagreement .
in fact , @xmath72 has a much more significant mean discrepancy @xmath142 in the @xmath89 mpc aperture .
in contrast , the @xmath71 have means that are entirely consistent for the ocdm and @xmath11 clusters in both apertures .
the variances of the centroided @xmath71 are significantly discrepant , particularly in the @xmath96 mpc aperture , where the ocdm variances exceed the @xmath11 variances .
this suggests a lower @xmath5 or @xmath6 is needed to bring the variances of the ocdm models into agreement with the @xmath11 sample .
the @xmath71 means of the lcdm clusters systematically exceed the @xmath11 means , but at a lower level of significance than does scdm .
the discrepancies are only significant in the @xmath89 mpc aperture , where @xmath73 @xmath143 and @xmath76 @xmath144 show the most significant discrepancies ; the even @xmath71 show at best a marginal discrepancy in their means @xmath145 .
the variances for the even @xmath71 are also significantly different , though only in the @xmath89 mpc aperture as well .
as the lcdm and scdm variances are very similar , we expect that the variance differences can be largely obviated with a lower value of @xmath5 .
the difference in the means of @xmath73 for the lcdm and @xmath11 clusters in the @xmath89 mpc aperture , though formally significant at better than the @xmath146 level , represents a shift of about one - half a decade in @xmath73 ; also , when using all 59 clusters of btb the significance is only @xmath147 ( @xmath134 ) . as we have discussed earlier , it is difficult to completely account for a discrepancy of this magnitude by invoking , e.g. , the unsuitability of the @xmath148 approximation , observational noise , or cooling flows .
we may make a more precise estimate of the effects of cooling flows on the @xmath71 .
the rosat clusters in the augmented edge sample all have estimated mass - flow rates ( fabian 1994 ) from which we may compute a luminosity ( bolometric ) due to the cooling flow following edge ( 1989 ) , @xmath149 erg / s , where @xmath150 is in @xmath151/year and @xmath152 is in kev . comparing this cooling luminosity to the total cluster luminosity , @xmath153 , using the results of david et al .
( 1993 ) allows us to in effect remove the cooling gas from the rosat @xmath71 . to a first approximation the cooling flow affects only @xmath66 because the cooling emission is weighted heavily towards the aperture center
hence , to approximately remove the effects of the cooling flows from the rosat clusters we reduce @xmath66 for each cluster by @xmath154 .
we find that the @xmath71 of the rosat clusters are modified minimally , the effect being that the means of the @xmath71 are increased by @xmath155 of a decade : means for @xmath72 and @xmath73 are -5.60 and -7.52 respectively in the @xmath89 aperture ; the variances show no significant systematic effect .
these small mean shifts do reduce the significance of the lcdm - rosat discrepancy , but the discrepancy is still significant at the @xmath133 level ; e.g. , @xmath156 for @xmath73 and @xmath157 for @xmath76 in the @xmath89 aperture , and @xmath158 for the even @xmath71 .
although cooling flows alone can not completely account for the differences in the rosat clusters and the lcdm model , it is very possible that when combined with the the other effects mentioned above a sizeable fraction of the half - decade difference could be made up which would in any event reduce the significance level of the difference . as a result
, we believe the discrepancy of the lcdm-@xmath11 means must be considered preliminary and await confirmation from appropriate hydrodynamical simulations .. ] on the other hand , the means of @xmath73 for the scdm and bcdm models exceed the @xmath11 means by almost a full decade to a higher formal significance level @xmath12 , which in light of the previous discussion should be considered robust . _
we conclude that the @xmath14 , cdm models can not produce the observed @xmath71 of the @xmath11 clusters , and that the discrepancy in @xmath71 means is due to @xmath0 being too large .
_ this agrees with our conclusions obtained in tb for the small sample of clusters drawn from the hydrodynamic simulation of navarro et al .
( 1995a ) . computed for the @xmath14 n - body simulations in this paper and the @xmath71 that tb computed for the navarro et al .
simulation further strengthens the scdm-@xmath11 discrepancy . ]
our conclusions are opposite those of mohr et al .
( 1995 ) who instead concluded that their _
cluster sample favored scdm over both ocdm and lcdm .
given the qualitative agreement discussed above between the jing et al . and mohr et al .
simulations , as well as between our present simulations and tb , it would seem that the discrepancy lies not in the details of the individual simulations .
moreover , since the centroid shift is qualitatively similar to our @xmath75 , and the axial ratio is qualitatively related to our @xmath159 , it would seem unlikely that we would reach entirely opposite conclusions .
the other plausible variable is to consider how btb and mohr et al . computed their statistics on the real cluster data .
the @xmath11 data analyzed by btb have better spatial resolution and sensitivity than the _ einstein _
data analyzed by mohr et al .. this implies that the mohr et al . data should be biased in the direction of less `` substructure '' with respect to btb , which is the opposite of what is found .
another important difference between the two investigations is that the @xmath71 are computed within apertures of fixed metric size , whereas mohr et .
use a @xmath160 criterion to define the aperture size . the fixed metric radius used by the @xmath71
ensures that cluster structure on the scale @xmath161 is compared consistently which is not true for the @xmath160 criterion ( see bta ) ; e.g. , mohr et al .
use aperture sizes of @xmath162 mpc for coma and of @xmath163 mpc for a2256 .
it is not obvious , however , how this confusion of cluster scales would explain the discrepancy of our results with mohr et al .. on the _ einstein _ sample of mohr et al .
, however such a task is beyond the scope of the present paper . ]
using the power ratios ( @xmath164 ) of buote & tsai ( 1995 bta ; 1996 btb ; tsai & buote 1996 tb ) we have examined the sensitivity of galaxy cluster morphologies to @xmath0 and @xmath1 using large , high - resolution n - body simulations .
x - ray images are generated from the dark matter by letting the gas density trace the dark matter .
we argue that the @xmath71 should not be seriously biased by this approximation because a real gas in hydrostatic equilibrium with potentials of cdm clusters is rounder , but also has a larger core radius , the effects of which partially cancel .
we also argue that the approximation should be reasonable during mergers because of the agreement shown between the evolution of the dark matter and gas found by tb who analyzed the hydrodynamical simulation of navarro et al .
( 1995a ) .
finally , the @xmath71 generated from the n - body simulations in this paper agree with results from the navarro et al .
hydrodynamical simulation ( tb ) .
similar agreement is seen between the results of the n - body simulations of jing et al .
( 1995 ) and the hydrodynamical simulations of mohr et al .
( 1995 ) . from analysis of several variants of the standard cold dark matter model
, we have shown that the @xmath71 can distinguish between models with different @xmath0 and @xmath1 .
generally , @xmath0 influences the means of the @xmath71 distributions such that larger values of @xmath0 primarily imply larger average pr values .
the slope of the power spectrum and @xmath5 primarily influence the variances of the @xmath71 ; smaller @xmath6 and @xmath5 generally imply smaller @xmath71 variances . for examining @xmath0
, our analysis indicates that @xmath73 is the best @xmath71 since its mean is quite sensitive to @xmath0 but very insensitive to @xmath1 .
it is advantageous also to consider @xmath72 when a cosmological constant is introduced since its means differ for the scdm and lcdm models by @xmath133 whereas @xmath73 only distinguishes the models at the @xmath134 level .
the dependence of the mean of @xmath72 on @xmath1 is not overly serious for studying differences in @xmath0 because the differences in means due to @xmath1 are always accompanied by larger differences in the variances ; i.e. different means but consistent variances mostly reflect differences in @xmath0 for @xmath72 . typically , the best apertures for segregating models are @xmath97 mpc
. we did not find it advantageous to compare the distributions along and perpendicular to the `` evolutionary tracks '' in the @xmath165 and @xmath166 planes ( see btb and tb ) .
the distributions along the tracks performed essentially as a weighted sum of the constituent @xmath71 .
the distributions perpendicular to the tracks were in almost all cases consistent for the models .
hence , although the evolutionary tracks are useful for categorizing the dynamical states of clusters , they do not allow more interesting constraints on @xmath0 and @xmath1 to be obtained over the individual @xmath71 .
the consistency of the distributions perpendicular to the evolutionary tracks seems to be a generic feature of the cdm models .
we compared the @xmath71 of the cdm models to the @xmath11 cluster sample of buote & tsai ( 1996 ) .
we find that the means of the @xmath167 ocdm and @xmath11 clusters are consistent , but the means of @xmath73 for the lcdm and @xmath11 clusters are formally inconsistent at the @xmath168 level .
we assert that this discrepancy should be considered marginal due to various issues associated with the simulation observation comparison . however , the means of @xmath73 for the scdm and bcdm models ( with @xmath14 ) exceed the @xmath11 means by almost a full decade with a high level of significance @xmath12 .
though the formal significance level of this @xmath13 / x - ray comparison should be considered only an approximation , we argue that taking into account the hydrodynamics and cooling will not reconcile a discrepancy this large . _
we conclude that the @xmath14 cdm models can not produce the observed @xmath71 of the @xmath11 clusters , and that the discrepancy in @xmath71 means is due to @xmath0 being too large .
_ this agrees with our conclusions obtained in tb for the small sample of clusters drawn from the hydrodynamic simulation of navarro et al .
( 1995a ) .
these conclusions are also consistent with other indicators of a low value of @xmath0 such as the dynamical analyses of clusters ( e.g. , carlberg et al . 1995 ) , the large baryon fractions in clusters ( e.g. , white et al .
1993 ) , and the heating of galactic disks ( toth & ostriker 1992 ) .
our conclusions are inconsistent with those of mohr et al .
( 1995 ) who instead concluded that their _
cluster sample favored @xmath14 , cdm over equivalents of our low - density models , ocdm and lcdm .
we argue that this type of discrepancy is unlikely due to numerical differences between our simulations .
we discuss possible differences due to how btb and mohr et .
al . computed their statistics on the real cluster data .
large hydrodynamical simulations are necessary to render the comparison to the @xmath11 data more robust . in addition , the effects of combining data at different redshifts needs to be explored since cluster formation rates should behave differently as a function of @xmath169 in different models ( e.g. , richstone et .
al . 1992 )
. it may also prove useful to apply @xmath71 to mass maps of clusters obtained from weak lensing ( kaiser & squires 1993 ) ) can distinguish between low - density and critical density models .
] , though for cosmological purposes it is not clear whether @xmath170 will be as responsive as @xmath171 to different @xmath0 and @xmath1 .
we gratefully acknowledge j. tsai for his role in facilitating this collaboration .
we thank e. bertschinger for suggesting to us the scheme to normalize the scale - free models using @xmath28 , t. beers for providing his rostat programs to compute ai and ti , and a. edge for providing the expression for @xmath172 .
we appreciated the anonymous referee s prompt reviewing and comments that helped improve the presentation of the paper .
dab was supported by grants nasgw-2681 ( through subcontract svsv2 - 62002 from the smithsonian astrophysical observatory ) and nag5 - 2921 , and acknowledges the hospitality of the institute of astronomy where the final stages of this work were carried out .
dab also expresses gratitude to several senior scientists who offered encouragement during the early stages of this project .
gx acknowledges support from nfs hpcc grant asc93 - 18185 and thanks the pittsburgh supercomputer center for use of the cray - t3d machine .
lccccccc + name & @xmath0 & @xmath27 & @xmath6 & @xmath5 & @xmath22 & @xmath173 + scdm & 1 & 0 & 1 & 1.00 & 0.5 & 20 + ocdm & 0.35 & 0 & 1 & 0.79 & 0.7 & 25 + lcdm & 0.35 & 0.65 & 1 & 0.83 & 0.7 & 39 + bcdm & 1 & 0 & 1 & 0.51 & 0.5 & 20 + sf00 & 1 & 0 & 0 & @xmath174 & @xmath174 & @xmath174 + sf10 & 1 & 0 & -1.0 & @xmath174 & @xmath174 & @xmath174 + sf15 & 1 & 0 & -1.5 & @xmath174 & @xmath174 & @xmath174 + sf20 & 1 & 0 & -2.0 & @xmath174 & @xmath174 & @xmath174 + lrc|rc|rc|rc|rc|rc + & & + + & & & & & & + & avg & @xmath175 & avg & @xmath175 & avg & @xmath175 & avg & @xmath175 & avg & @xmath175 & avg & @xmath175 + + + scdm & -5.14 & 0.86 & -5.16 & 0.83 & -5.38 & 0.76 & -6.72 & 0.73 & -6.82 & 1.01 & -7.06 & 0.98 + ocdm & -5.55 & 0.61 & -5.82 & 0.78 & -5.93 & 1.03 & -7.40 & 0.81 & -7.56 & 1.13 & -7.59 & 1.28 + lcdm & -5.45 & 0.83 & -5.69 & 0.85 & -5.87 & 0.83 & -7.08 & 0.83 & -7.32 & 1.00 & -7.38 & 1.02 + bcdm & -5.01 & 0.57 & -5.24 & 0.60 & -5.41 & 0.63 & -6.83 & 0.65 & -6.98 & 0.67 & -7.07 & 0.68 + sf00 & -5.34 & 0.82 & -5.64 & 1.00 & -5.75 & 0.97 & -7.12 & 1.02 & -7.28 & 1.14 & -7.35 & 1.10 + sf10 & -5.02 & 0.57 & -5.22 & 0.71 & -5.47 & 0.85 & -6.89 & 0.87 & -7.09 & 1.02 & -7.36 & 1.18 + sf15 & -5.20 & 0.85 & -5.22 & 0.94 & -5.40 & 0.91 & -6.92 & 0.99 & -6.93 & 1.02 & -7.05 & 0.96 + sf20 & -5.07 & 0.60 & -5.24 & 0.62 & -5.45 & 0.57 & -7.01 & 0.71 & -7.13 & 0.75 & -7.20 & 0.79 + ccc|rrr|rrr|rrr + & & & & & + & & & & & & & & & & & + & ( % ) & ( % ) & ( % ) & ( % ) & ( % ) & ( % ) & ( % ) & ( % ) & ( % ) + + + + + scdm & vs. & ocdm & 1.63 & 4.09 & 0.94 & 0.06 & 71.71 & 0.07 & 0.89 & 7.18 & 0.94 + scdm & vs. & lcdm & 10.47 & 87.90 & 21.79 & 0.74 & 89.05 & 1.97 & 0.83 & 58.78 & 3.90 + ocdm & vs. & lcdm & 54.64 & 5.79 & 51.42 & 47.52 & 61.72 & 21.79 & 78.26 & 20.53 & 34.56 + scdm & vs. & bcdm & 43.06 & 1.51 & 98.09 & 65.19 & 4.65 & 21.79 & 83.52 & 24.87 & 34.56 + sf00 & vs. & sf20 & 10.76 & 6.35 & 7.31 & 4.19 & 0.38 & 0.94 & 10.44 & 0.13 & 3.90 + + + + + scdm & vs. & ocdm & 0.02 & 52.99 & 0.18 & 0.31 & 46.71 & 0.94 & 4.43 & 10.64 & 7.31 + scdm & vs. & lcdm & 4.44 & 43.91 & 21.79 & 3.26 & 97.10 & 7.31 & 15.88 & 81.58 & 51.42 + ocdm & vs. & lcdm & 9.10 & 88.39 & 21.79 & 30.99 & 44.52 & 51.42 & 43.51 & 16.61 & 51.42 + scdm & vs. & bcdm & 49.58 & 47.26 & 70.85 & 40.62 & 1.39 & 34.56 & 97.46 & 2.48 & 70.85 + sf00 & vs. & sf20 & 57.65 & 2.89 & 21.79 & 49.89 & 1.28 & 21.79 & 50.92 & 4.31 & 12.97 + llll|lll + & & + models & @xmath176 & @xmath177 & @xmath178 & @xmath176 & @xmath177 & @xmath178 + + + + + scdm & 0.60e-01 & 0.12e-01 & 0.68e-03 & 0.14e-01 & 0.24e+01 & 0.33e-01 + bcdm & 0.10e-04 & 0.12e+02 & 0.56e-04 & 0.12e-01 & 0.20e+02 & 0.12e-01 + ocdm & 0.23e+02 & 0.52e+01 & 0.16e+02 & 0.68e+02 & 0.23e-01 & 0.43e+02 + lcdm & 0.11e+02 & 0.21e-01 & 0.13e+01 & 0.41e+02 & 0.69e+00 & 0.23e+02 + + + + + scdm & 0.15e-03 & 0.71e+02 & 0.69e-01 & 0.17e+01 & 0.20e+02 & 0.10e+02 + bcdm & 0.64e-03 & 0.28e+02 & 0.32e-02 & 0.35e+00 & 0.46e+02 & 0.12e+02 + ocdm & 0.22e+02 & 0.80e+02 & 0.42e+02 & 0.93e+02 & 0.83e+00 & 0.29e+02 + lcdm & 0.44e+00 & 0.69e+02 & 0.97e+00 & 0.33e+02 & 0.14e+02 & 0.32e+02 + contour plots of the `` x - ray images '' for 16 of the 39 clusters analyzed in the scdm model obtained from projecting @xmath180 . each image is @xmath46 @xmath47 mpc@xmath48 and the axes units are @xmath49 mpc pixels . the standard deviation as a function of the average value of the prs in the @xmath98 mpc aperture for the models in [ omega ] .
the error bars represent @xmath181 errors estimated from 1000 bootstrap resamplings . the standard deviation as a function of the average value of the prs in the @xmath98 mpc aperture for the models in [ spectrum ] .
the error bars represent @xmath181 errors estimated from 1000 bootstrap resamplings . the standard deviation as a function of the average value of the prs in the @xmath89 mpc aperture for the rosat clusters and the models discussed in [ rosat ] .
the error bars represent @xmath181 errors estimated from 1000 bootstrap resamplings . | we examine the sensitivity of the spatial morphologies of galaxy clusters to @xmath0 and @xmath1 using high - resolution n - body simulations with large dynamic range .
variants of the standard cdm model are considered having different spatial curvatures , scdm @xmath2 , ocdm @xmath3 , lcdm @xmath4 , and different normalizations , @xmath5 .
we also explore critical density models with different spectral indices , @xmath6 , of the scale - free power spectrum , @xmath7 .
cluster x - ray morphologies are quantified with power ratios ( prs ) , where we take for the x - ray emissivity @xmath8 , which we argue is a suitable approximation for analysis of prs .
we find that @xmath0 primarily influences the means of the pr distributions whereas the power spectrum ( @xmath6 and @xmath5 ) primarily affects their variances : @xmath9 is the cleanest probe of @xmath0 since its mean is very sensitive to @xmath0 but very insensitive to @xmath1 .
the pr means easily distinguish the scdm and ocdm models , while the scdm and lcdm means show a more modest , but significant , difference @xmath10 .
the ocdm and lcdm models are largely indistinguishable in terms of the prs .
finally , we compared these models to a sample of @xmath11 clusters and find that the pr means of the scdm clusters exceed the @xmath11 means with a high formal level of significance @xmath12 . though the formal significance level of this @xmath13 / x - ray comparison should be considered only approximate , we argue that taking into account the hydrodynamics and cooling will not reconcile a discrepancy this large .
the pr means of the ocdm clusters are consistent , and the means of the lcdm clusters are marginally consistent , with those of the @xmath11 clusters .
thus , we conclude that cluster morphologies strongly disfavor @xmath14 , cdm while favoring low density , cdm models with or without a cosmological constant . |
nowadays , it is generally accepted that quantum chromodynamics ( qcd ) is the theory of the strong interaction .
it has been very successful and tested to great precision at high energies ; however , its application in the low energy region of @xmath2 1 gev is quite problematic due to the large running coupling constant . the advent of chiral perturbation theory ( @xmath3pt ) and lattice qcd approach has made possible a model independent study of the low - energy strong phenomena for the first time .
neutrino physics has made remarkable progress in recent years , as evidenced by the 2002 nobel prize in physics ( awarded partly to raymond davis jr and masatoshi koshiba `` for pioneering contributions to astrophysics , in particular for the detection of cosmic neutrinos '' ) .
after many years of experimental ( and theoretical ) efforts , two facts have been firmly established : ( i ) neutrino have masses and ( ii ) different flavors of neutrino can oscillate into each other .
presently , one of the main goals in the field is to measure accurately the masses and oscillation parameters .
a good understanding of pion production is relevant to reduce systematic uncertainties in oscillation experiments .
the axial nucleon to @xmath1(1232 ) transition , characterized by four form factors , plays an important role in this reaction at low @xmath4 transfer.@xcite most of our current ( experimental ) knowledge of the @xmath0 axial transition form factors comes from neutrino bubble chamber data.@xcite the possibility to extract them using parity - violating electron scattering at jefferson lab has been extensively studied,@xcite and could shed new light on the nature of these form factors .
present and future neutrino experiments ( miniboone , k2k , fermilab ) could also provide further information . in the past , the theoretical descriptions have been done using different approaches , mostly quark models ( for a review , see ref . ) . in recent years , there has been an increasing interest on these form factors .
they have been calculated , for instance , using the chiral constituent quark model and light cone qcd sum rules .
state of the art calculations within lattice qcd have also become available.@xcite while the axial @xmath0 form factors have been addressed in ( tree level ) hb@xmath3pt,@xcite _ no calculation has been performed up to now within the relativistic framework_. with lattice qcd results becoming available and in view of the many ongoing experimental efforts to extract these form factors from electron- and neutrino - induced reactions , it is timely to study the axial @xmath0 transition form factors within covariant baryon @xmath3pt.@xcite
the study of the @xmath0 transition form factors using covariant baryon @xmath3pt is much more complicated than it seems to be . to begin with
, one has to address the following three questions : power counting , chiral lagrangians , and the appropriate form of the @xmath1 propagator . 1
. a proper power counting scheme is at the center of effective field theories . to include the @xmath5 explicitly
, one has to count the @xmath6-@xmath1 mass difference @xmath7gev properly . in the present work ,
we adopt the @xmath8 expansion scheme , which counts @xmath9 as @xmath10 to maintain the scale hierarchy @xmath11.@xcite 2 .
the pion - nucleon and pion - pion lagrangians are rather standard .
the @xmath12 and @xmath13 lagrangians , on the other hand , require more attention .
the @xmath5 is a spin-3/2 resonance and , therefore , its spin content can be described in terms of the rarita - schwinger ( rs ) field , @xmath14 , where @xmath15 is the lorentz index .
this field , however , contains unphysical spin-1/2 components . in order to tackle this problem
, we follow ref . and adopt the `` consistent '' couplings , which are gauge - invariant under the transformation @xmath16 3 .
different forms of the spin-3/2 propagator have been used in the literature , some of which may lead to serious theoretical problems.@xcite due to the spin-3/2 gauge symmetric nature of the consistent couplings , we can use the most general spin-3/2 free field propagator.@xcite a more detailed discussion of these issues and the relevant @xmath12 and @xmath13 lagrangians can be found in ref . .
axial transition form factors up to order @xmath17 .
the double , solid , and dashed lines correspond to the delta , nucleon , and pion , respectively ; while the wiggly line denotes the external pseudovector source.[fig_diagram ] ] the @xmath0 axial transition form factors can be parametrized in terms of the usually called adler form factors:@xcite @xmath18 where @xmath19 is the third isospin component of the axial current . in the @xmath8 expansion scheme , up to order @xmath17 , all diagrams contributing to the axial @xmath0 transition form factors
are displayed in fig .
[ fig_diagram ] .
the corresponding results in terms of low - energy constants ( lec ) and loop functions are summarized in table [ table : avf ] .
we can easily see that at @xmath20 , @xmath21 , where @xmath22 is the @xmath23 coupling determined from the @xmath1 width , which is close to the kitagaki - adler value@xcite of 1.2 .
the kitagaki - adlder assumption @xmath24 , on the other hand , is satisfied only up to @xmath25 , i.e. , non pion - pole contributions appear at order @xmath17.@xcite the form factors @xmath26 and @xmath27 both start at chiral order 2 and get their @xmath28 dependence at order 3 from the loops . for @xmath26
, we find a small @xmath28 dependence , which is quite sensitive to the @xmath29 coupling constant . on the other hand , its imaginary part ,
coming mainly from the @xmath6-@xmath6 internal diagram , is finite ( @xmath30 at @xmath31 ) and has a mild @xmath28 dependence .
this suggest that @xmath26 is small ( compared to @xmath32 ) but not necessarily zero .
the @xmath27 dependence on @xmath28 is also found to be rather mild at order @xmath17 . in covariant
@xmath3pt up to @xmath17 , four @xmath25 and seven @xmath17 lec appear in the results . however , some of them appear in particular combinations .
therefore , effectively we have only five unknown constants
. they can be fixed by fitting either to the phenomenological form factors obtained from neutrino bubble chamber data ( with several assumptions ) , to the results of other approaches , such as those of various quark models , or to the lattice qcd results@xcite . for a more detailed discussion ,
see ref . .
we thank mauro napsuciale , stefan scherer , wolfram weise , and in particular massimiliano procura and vladimir pascalutsa for useful discussions .
l. s. geng acknowledges financial support from the ministerio de educacion y ciencia in the program `` estancias de doctores y tecnologos extranjeros '' .
j. martin camalich acknowledges the same institution for a fpu fellowship .
this work was partially supported by the mec contract fis2006 - 03438 , the generalitat valenciana acomp07/302 , and the eu integrated infrastructure initiative hadron physics project contract rii3-ct-2004 - 506078 .
l. alvarez - ruso , l. s. geng and m. j. vicente vacas , phys .
c * 76 * , 068501 ( 2007 ) .
t. kitagaki _ et al .
_ , phys .
d * 42 * , 1331 ( 1990 ) .
n. c. mukhopadhyay , m. j. ramsey - musolf , s. j. pollock , j. liu and h. w. hammer , nucl .
phys . a * 633 * , 481 ( 1998 ) .
j. liu , n. c. mukhopadhyay and l. s. zhang , phys . rev .
c * 52 * , 1630 ( 1995 ) . c. alexandrou , t. leontiou , j. w. negele and a. tsapalis , phys .
* 98 * , 052003 ( 2007 ) . s. l. zhu and m. j. ramsey - musolf , phys .
d * 66 * , 076008 ( 2002 ) .
l. s. geng , j. m. camalich , l. alvarez - ruso and m. j. vicente vacas , arxiv:0801.4495 [ hep - ph ] . v. pascalutsa and d. r. phillips , phys .
c * 68 * , 055205 ( 2003 ) .
v. pascalutsa , m. vanderhaeghen and s. n. yang , phys .
rept . * 437 * , 125 ( 2007 ) .
m. benmerrouche , r. m. davidson and n. c. mukhopadhyay , phys .
c * 39 * , 2339 ( 1989 ) .
v. pascalutsa and m. vanderhaeghen , phys .
b * 636 * , 31 ( 2006 ) . | we have performed a theoretical study of the axial nucleon to delta(1232 ) ( @xmath0 ) transition form factors up to one - loop order in covariant baryon chiral perturbation theory within a formalism in which the unphysical spin-1/2 components of the @xmath1 fields are decoupled . |
tight binding models , such as the hubbard hamiltonian , have been widely studied for their ability to describe different kinds of insulating phases present in condensed matter systems .
interactions between fermions give rise to gapped mott insulators where the fermions are immobile at commensurate fillings .
for bipartite lattices , mott behavior is typically associated with antiferromagnetic order . on the other hand ,
the imposition of an external periodic potential can drive band insulating behavior . in systems where several types of insulating effects are present
, phase transitions can take place as the relative strength of the parameters is changed .
close to such transitions , the competition between these different terms allows smaller effects , such as kinetic energies , to result in low temperature metallic phases or more exotic types of correlations such as bond - order .
such compensating effects are believed to explain enhanced response in different quasi - one - dimensional systems such as organic materials @xcite , ferroelectric perovskites @xcite or transition metal - oxides @xcite .
a simple model exhibiting this kind of behavior is the ionic hubbard hamiltonian @xcite , a tight binding model that includes an on - site repulsion and a staggered potential that takes alternating values on neighboring sites of a bipartite lattice . at half - filling
, the on - site repulsion drives the system into an antiferromagnetic mott insulator ( mi ) , with a homogeneous density . when the staggered potential dominates , a band insulator ( bi ) with a regular modulation of the density results .
a quantum phase transition between these two states occurs in the phase diagram . in one dimension
, the ionic hubbard model has been widely studied through a variety of techniques including density matrix renormalization group ( dmrg ) , exact diagonalization , bosonization and quantum monte carlo ( qmc ) with on - site @xcite or extended @xcite interactions .
the precise conclusions are still subject to some debate , but many of the studies suggest that the direct transition from the band insulator to the mott insulator is replaced by an intervening insulating bond - ordered phase . in two dimensions , dynamical mean field theory ( dmft)@xcite , cluster - dmft@xcite and determinant quantum monte carlo ( dqmc ) @xcite were used . the existence of an intermediate phase between the bi and mi phases has been established , as in one dimension , but the precise nature of the phase is still under debate ( see section ii ) .
studies of a bilayer hubbard model @xcite and in infinite dimension @xcite yield similar results with a low temperature metal as the intermediate phase . here the interplanar coupling , if sufficiently large , gives a noninteracting band structure with a gap at half - filling much like a staggered potential .
a recent experiment attempts to observe this kind of phase @xcite .
a related problem is the study of heterostructures of srtio@xmath0 and latio@xmath0 , one being a mi , the other a bi and a metallic phase appearing at the interface between the two @xcite .
some of the phenomena presented here also have similarities with those found in corresponding disordered models used to describe binary alloys @xcite .
in this paper , we consider the different phases and transitions of the ionic hubbard hamiltonian on a two - dimensional square lattice using determinant quantum monte carlo ( dqmc ) , extending our initial study@xcite . in section
ii , we introduce the model and summarize our previous results . in section iii , we briefly describe the numerical techniques used while in section iv , we investigate further the different phases by measuring the charge and magnetic structure factors and the spectral function .
we then present an analysis of the properties away from half filling ( section v ) and the search for bond - order waves ( section vi ) .
conclusions are in section vii .
we consider the ionic hubbard model @xmath1 ( n_{{\bf r } \uparrow } + n_{{\bf r } \downarrow } ) \label{hamiltonian}\end{aligned}\ ] ] where @xmath2 are the usual fermion creation ( destruction ) operators on site @xmath3 of a two - dimensional square lattice with spin @xmath4 and where @xmath5 is the number operator .
the hopping parameter , @xmath6 , sets the energy scale and is fixed at a value of @xmath7 , while tuning the chemical potential @xmath8 gives the desired fermion filling .
the interaction strength @xmath9 and staggered potential @xmath10 are the two competing terms of our system , the former encouraging single occupancy of all sites and the latter double occupancy of the low energy sublattice.@xcite considering the limiting cases @xmath11 and @xmath12 we find two distinct insulating phases at half - filling ( density @xmath13 ) . for @xmath14 ,
the usual translationally invariant hubbard model arises and , at @xmath15 , forms an antiferromagnetic mott insulator at half - filling for any value of @xmath9 . at strong coupling ,
the gap is set by @xmath9 , while the effective anti - ferromagnetic coupling is set by @xmath16 .
these two energy scales are evident in the magnetic properties where @xmath17 demarks the onset of moment formation on individual sites and @xmath18 , the onset of antiferromagnetic order between sites . in the non interacting limit @xmath19
, the hamiltonian can be diagonalized in fourier space ( where it reduces to a set of independent @xmath20 matrices ) @xmath21 where the sum runs over half the brillouin zone and where @xmath22 is the energy of a free particle of wave vector @xmath23 .
the eigenenergies @xmath24 are then split into two bands separated by a gap @xmath25 and , in this limit , the system is a band insulator at half - filling .
these two insulating phases can be seen schematically in fig .
[ schem ] .
recently , it was shown@xcite that there is no direct transition between these two insulating phases , but that an intermediate phase appears ( fig .
the precise nature of the phase is still being debated . in previous work @xcite
, we used the determinant quantum monte carlo algorithm ( dqmc ) to study the conductivity of the model . performing a finite size and temperature analysis
, we concluded that the intermediate phase has a finite conductivity and is , consequently , metallic .
_ arrive at the same conclusion in their dynamical mean field theory ( dmft ) study@xcite while using a cluster - dmft analysis , kancharla _ et al .
_ argue for a bond ordered wave ( bow ) phase @xcite .
dqmc@xcite is a numerical approach which expresses the partition function @xmath26 as a path integral where the inverse temperature @xmath27 is discretized .
the interaction term is decoupled via a hubbard - stratonovich ( hs ) transformation , introducing a sum over an ising spin field and leaving the hamiltonian in a quadratic form with respect to the fermion operators .
the fermionic degrees of freedom can then be integrated out and the remaining boltzmann weight is expressed as the determinant of the product of two matrices that depend on the hs spin variables . in the conventional hubbard model ,
this determinant is positive at half - filling , due to particle - hole symmetry . with the staggered potential ,
the determinant is no longer always positive at half - filling resulting in a sign problem which can be severe .
the remaining sum over the hs field and the evaluation of observables is achieved by a metropolis monte carlo algorithm .
it is difficult to obtain good estimates for the observables when the determinant is often negative , _
i.e. _ , when the mean value of the sign of the determinant is small .
this problem is important for intermediate values of @xmath9 and @xmath25 , while it is not severe in the limiting cases where we retrieve simple bi or mi phases ( @xmath9 or @xmath25 small or large ) . with this formulation , space and (
imaginary-)time separated green functions @xmath28 are easily expressed as elements of the matrices .
> from the spatial fourier transform of @xmath29 , one can extract the spectral function ( essentially the density of states ) , by inverting the laplace transform @xmath30 solving for @xmath31 remains a non - trivial task that is performed with a method of analytic continuation@xcite .
we measure correlation functions for charge and spin density waves ( cdw and sdw , respectively ) , given by @xmath32 \left[n_{{\bf r } \uparrow}\pm n_{{\bf r } \downarrow}\right]\right\rangle\nonumber\ ] ] where @xmath33 corresponds to cdw and @xmath34 to sdw .
the staggered potential imposes a checkerboard modulation on the density , not only in the bi phase but in all phases as it breaks explicitly the translational invariance .
antiferromagnetic magnetic correlations are also expected in the mi phase .
thus , for both types of correlations , we expect ordering with a wave vector @xmath35 , characterized by the divergence of corresponding structure factors @xmath36
to complete our previous analysis@xcite of the intermediate phase shown in fig .
[ pd ] we evaluated the charge and spin correlations and spectral function .
when the size is not explicitly mentioned , we used a @xmath37 sites lattice . @xmath38 denotes the side of the lattice .
[ spectralfunc ] shows the spectral function for @xmath39 ( along the vertical line shown in fig .
[ pd ] ) at @xmath40 and for three values of @xmath9 corresponding to the three different phases . for @xmath41 and @xmath42 , corresponding to the bi and mi phases , the spectral function shows a gap around @xmath43 , thus indicating an insulator .
in contrast , the spectral function for @xmath44 is non zero and shows a quasi - particle peak at @xmath43 which indicates metallic behavior . these results are consistent with measurements of the conductivity in our previous article , as is confirmed by the direct comparison of the spectral function at zero frequency @xmath45 and the conductivity @xmath46 shown in fig . [ specvsu ] . for @xmath47 and as @xmath9
is increased , the behaviors of these two quantities are similar : they both show , for intermediate values of @xmath9 , a peak that grows with @xmath27 .
the ranges of interaction where there are peaks are slightly different because of finite size effects .
@xmath48 and @xmath49 are the critical repulsion energies that cause bi to metal and metal to mi phase transitions for a fixed value of @xmath25 .
these critical values are obtained by a finite temperature analysis of the conductivity@xcite .
as the temperature is lowered , the conductivity increases in a metal and decreases in an insulator . locating the points where the behavior of the conductivity changes gives the critical values . for @xmath47 ,
the bi - metal transition is located at @xmath50 , the metal - mi transition is located at @xmath51 and the metallic region is in between .
structure factors @xmath52 and @xmath53 are shown in fig .
[ structfac ] for @xmath25= 0.25 and for inverse temperatures @xmath54 and 16 .
for these values of @xmath25 , the metallic boundaries are @xmath55 and @xmath56 .
as @xmath9 is increased , @xmath52 increases as the system is driven towards the mi phase , while @xmath53 decreases , in moving away from the bi phase .
the transitions in these quantities when we cross the phase boundaries are not sharp on lattices of fixed , finite size owing to finite size rounding effects .
in addition , as stated above , a smoother evolution is expected for @xmath53 as a modulation of @xmath57 is imposed by the external potential for all parameter values . in a conventional hubbard model ,
the mi at half filling is always an antiferromagnet .
however , it has been suggested recently for related models @xcite that the metal - mi and para- to antiferro - magnetic ( pm - af ) transitions may not take place together . a finite size scaling analysis of @xmath52 for different values of @xmath9 is needed to check whether these two boundaries coincide in our model .
the structure factor @xmath52 increases like the number of sites @xmath58 in the thermodynamic limit @xmath59 where @xmath60 is the antiferromagnetic order parameter .
[ finitesizeaf ] shows structure factors for sizes up to @xmath61 sites for different values of @xmath9 .
@xmath25 is set to 0.5 and the metallic phase appears between @xmath62 and @xmath63 .
we apply a least squares fit to the data , and extrapolate to an infinite system .
we find a non - vanishing structure factor only in the mi phase ( @xmath64 and @xmath65 ) . in all the other cases ,
including points located close to the metal - mi transition ( @xmath66 and @xmath67 ) , the structure factor goes to zero as the size is increased .
the pm - af transition , therefore , takes place between @xmath68 and @xmath64 and appears to be simultaneous with the metal - mi transition .
although our study on limited sizes can not completely exclude completely the possibility of an exotic af metallic phase located around the metal - mi boundary , our data suggest that it does not occur .
we extend our study to determine the behavior away from half - filled band . in fig .
[ rhovmu1 ] , the density @xmath57 is plotted as a function of chemical potential .
the compressibility @xmath69 is given by @xmath70 .
for @xmath69=0 , incompressible insulating phases are present which is indicated by plateaux in the @xmath57 versus @xmath8 curve .
both the mi ( @xmath71 ) and bi ( @xmath72 ) phases have this distinctive plateau at half - filling .
the metallic nature of the intermediate phase is once again confirmed by the absence of a plateau around @xmath13 . using the same parameters @xmath9 and @xmath25
, we show in fig .
[ structvrho ] the evolution of the structure factors as functions of @xmath57 .
as expected , for the parameters corresponding to the mi at @xmath13 , @xmath52 shows a strong peak around half - filling indicating the presence of the af - mi in this case and its rapid suppression upon doping . in this case
, charge correlations always remain small , as double occupancy of sites is essentially forbidden .
similarly , for the parameters corresponding to the bi , the peak is present in @xmath53 at @xmath13 and @xmath52 remains small for all densities , as the occupation of larger energy site is suppressed . for the metallic case
, there is no peak in the magnetic structure factor at @xmath13 but a moderate peak in @xmath53 .
in all these three cases , away from half - filling , the structure factors are rapidly suppressed and the system is compressible and becomes a paramagnetic metal . for stronger interactions and staggered potential ,
other ordered phases can be found at quarter and three quarters fillings . for @xmath42 and @xmath73 ,
there is a band insulator at half - filling shown by the @xmath13 plateau in fig .
[ rhovmu2 ] .
additional plateaux emerge at @xmath74 and @xmath75 .
the plateau at @xmath76 corresponds to a state where all the lower energy sites are occupied by one fermion . adding a fermion to an already occupied site increases the energy approximately by @xmath9
, while adding it to an empty ( high energy ) site increases it by @xmath25 .
there is then an energy gap , roughly equal to the smaller of the two parameters ( @xmath9 in this case ) .
we then have a mott insulator at quarter filling and , symetrically , another one at three quarters filling . in the case where @xmath9 is larger than @xmath25 , the situation is reversed and we expect a mi at half - filling and a bi for @xmath76 and @xmath77 ( this case is difficult to treat numerically because of a strong sign problem ) .
although the system is incompressible for @xmath76 , there is no concommitant magnetic order : in fig .
[ structvmu ] , @xmath52 always remains small as @xmath57 is varied away from half - filling and there is no af peak around quarter filling ( the maximum being around @xmath78 ) .
the af ordering appears when the fermions jump between neighboring sites , thus creating an effective antiferromagnetic coupling between spins . at quarter filling , the particles are surrounded by empty sites and there is no effective coupling . as for @xmath53 we only observe a peak at half - filling ( fig . [ structvmu ] ) . a plot of the conductivity ( fig .
[ structvmu](c ) ) versus @xmath57 shows that , indeed , @xmath46 vanishes for @xmath79 and @xmath80 corresponding to the plateaux observed in fig . [ rhovmu2 ] .
long range bond - ordered waves ( bow ) are known to exist in one dimension in the ionic hubbard model @xcite as well as in other one - dimensional systems exhibiting density modulation such as extended hubbard models @xcite .
( in the latter case , the two numerical treatments , using the stochastic series expansion@xcite and density matrix renormalization group@xcite , disagree on the precise width of the bow phase as well as over what range of coupling values it is present . )
it was suggested by kancharla _
et al._@xcite that the intermediate phase observed at half - filling in the two - dimensional case studied here was in fact a bond - ordered state , as in one dimension .
other studies have attempted to observe bond - order in two dimensions , for example in extended hubbard models @xcite . in two dimensions , bond orders
have mostly been obtained in hamiltonians possessing rather special geometries or interactions , e.g. on checkerboard lattices @xcite or by including additional terms such as ring exchanges @xcite .
hence , bond order seems in general more difficult to obtain in two dimensions than in one . in our system
, we can understand this with the argument sketched in fig .
[ bowschem ] . starting from a situation where all the particles are located on the low energy sites ( @xmath81 sites ) ,
a jump to a high energy site ( @xmath82 site ) increases the energy roughly by @xmath83 .
if @xmath25 and @xmath9 are of the same order of magnitude , this energy cost is small and the jump has a high probability to happen .
this @xmath82 site is connected to two @xmath81 sites : the one where the jumping particle comes from ( to the right in fig .
[ bowschem ] ) and another one ( to the left ) still occupied by two particles .
the jumps on the left bond are suppressed : the particle on the @xmath82 site and the particle having the same spin ( an up spin in the sketch ) on the @xmath81 site can not jump because of pauli principle , while a jump of the remaining particle has a large energy cost @xmath25 . in short , having a particle jumping on a bond ( and thus a contribution to the kinetic energy on this bond ) will inhibit jumps on the following bond . in one dimension , this can lead to a bond order with alternate strong and weak values of the kinetic energy . on a two dimensional square lattice
, the appearance of a strong bond will inhibit the three other bonds linked to the same site and it seems more difficult to generate regular arrays of weak and strong bonds . to check if there is a bow , we examined the correlation function of the kinetic energy @xmath84 where @xmath85 is the kinetic energy operator connecting sites @xmath86 and @xmath87 ( @xmath88 being a unit vector directed along the @xmath89 direction ) , @xmath90 the fourier transform of @xmath91 always exhibits a very small peak for @xmath92 , in _ _ all the different phases at or away from half - filling .
finite size scaling analysis indicates that these peaks persist in large size lattices , but this is the case for all the different phases , not only in the intermediate phase .
we also studied the bow susceptibility@xcite for this wave vector @xmath93 fig .
[ corrbow ] shows the evolution of @xmath94 at half - filling as a function of @xmath9 and for two different values of @xmath25 .
the curve is maximum in the intermediate phase but is also non zero in other phases .
similar results are obtained for bonds oriented along the @xmath95 direction and for mixed @xmath89-@xmath95 cases .
we do not believe these signals to indicate the presence of a bow phase for the following reasons ( a ) the values obtained are extremely small .
( b ) these `` bow signals '' are ubiquitous in _ all three phases _ and not specific to the intermediate one .
the susceptibility is , for example , nearly as important in the mi phase as in the intermediate phase .
as soon as @xmath9 and @xmath25 are both non - zero , we observe this very weak non - zero value of @xmath94 .
( c ) starting from the metallic phase ( @xmath96 ) and doping away from half - filling the system is certainly known to be metallic ( see fig . [ corrbow2 ] ) yet the bow susceptibility is non zero and comparable to the value found at half - filling .
for all these reasons , we believe that these tiny signals are only a weak signature of an interplay between the interaction and externally imposed checkerboard potential .
we have extended our previous study of the phase diagram of the ionic hubbard model @xcite .
in addition to new qmc results for the conductivity , we calculated the spectral function , the kinetic energy correlation function ( the order parameter for the bow phase ) and the magnetic and charge order parameters .
our results point to the presence of a normal conducting metallic phase intervening between the band and mott insulating phases .
in particular , we have not found a clear signal in the bow order parameter and susceptibilities indicating the presence of such a phase .
we also studied the system doped away from half filling .
when the band or mott insulating phase is doped , a conducting phase is obtained except at fillings of @xmath97 and @xmath98 where we found insulating phases without magnetic order .
t. wilkens and r.m .
martin , phys .
b * 63 * , 235108 ( 2001 ) ; m.e .
torio , a.a .
aligia , and h.a .
ceccatto , phys .
b * 64 * , 121105(r ) ( 2001 ) ; c.d .
batista and a.a .
aligia , phys .
lett . * 92 * , 246405 ( 2004 ) ; g.i .
japaridze , r. hayn , p. lombardo , and e. mller - hartmann , cond - mat/0611415 ( 2006 ) .
our previous article @xcite contained a factor of two error in the value of the staggered potential @xmath25 .
the correct definition of the hamiltonian is the one given here . the bi - mi transition line in the @xmath99 limit in fig .
[ pd ] has been modified accordingly .
k. byczuk , w. hofstetter , and d. vollhardt , phys .
b * 69 * , 045112 ( 2004 ) ; p.
lombardo , r. hayn , and g.i .
japaridze , phys .
b * 74 * , 085116 ( 2006 ) ; n. paris , a. baldwin , and r.t .
scalettar , cond - mat/0607427 ( 2006 ) . | we investigate the phases of the ionic hubbard model in a two - dimensional square lattice using determinant quantum monte carlo ( dqmc ) . at half - filling , when the interaction strength or the staggered potential dominate we find mott and band insulators , respectively .
when these two energies are of the same order we find a metallic region .
charge and magnetic structure factors demonstrate the presence of antiferromagnetism only in the mott region , although the externally imposed density modulation is present everywhere in the phase diagram .
away from half - filling , other insulating phases are found .
kinetic energy correlations do not give clear signals for the existence of a bond - ordered phase . |
the inner and outer tensor products of irreducible characters of the symmetric groups ( or equivalently of schur functions ) have been of central interest in representation theory and algebraic combinatorics since the landmark papers of littlewood and richardson @xcite and murnaghan @xcite .
more recently , these coefficients have provided the centrepiece of geometric complexity theory ( an approach that seeks to settle the @xmath0 versus @xmath1 problem @xcite ) and have been found to have deep connections with quantum information theory @xcite .
the coefficients arising in the outer tensor product are the most well - understood . the littlewood
richardson rule provides an efficient positive combinatorial description for their computation . using this algorithm ,
a classification of multiplicity - free outer tensor products was obtained by stembridge @xcite .
this was extended to a classification of multiplicity - free skew characters by gutschwager @xcite , a result equivalent to the classification of multiplicity - free products of schubert classes obtained around the same time by thomas and yong @xcite . by contrast , the coefficients arising in the inner tensor product are much less well - understood ; indeed , they have been described as ` perhaps the most challenging , deep and mysterious objects in algebraic combinatorics ' @xcite . the determination of these coefficients has been described by richard stanley as ` one of the main problems in the combinatorial representation theory of the symmetric group ' @xcite . while ` no satisfactory answer to this question is known ' @xcite there have , over many decades , been a number of contributions made towards computing special products ( such as those labelled by 2-line or hook partitions @xcite ) or the multiplicity of special constituents ( for example those with few homogenous components @xcite ) . in 1999 , bessenrodt conjectured a classification of multiplicity - free kronecker products of irreducible characters of the symmetric groups . mainly using results of remmel , saxl and vallejo ,
it was shown at that time that the products on the conjectured list were indeed multiplicity - free and the conjecture was verified by computer calculations for all @xmath2 .
since then , multiplicity - free kronecker products have been studied in @xcite . in this paper
we prove that the classification list is indeed complete for all @xmath3 and hence confirm the conjecture , that is , we have the following result : [ thm : classification ] let @xmath4 be partitions of @xmath3
. then the kronecker product @xmath5\cdot [ \mu]$ ] of the irreducible characters @xmath5,[\mu]$ ] of @xmath6 is multiplicity - free if and only if the partitions @xmath4 satisfy one of the following conditions ( up to conjugation of one or both of the partitions ) : 1 .
one of the partitions is @xmath7 , and the other one is arbitrary ; 2 .
one of the partitions is @xmath8 , and the other one is a fat hook ( here , a fat hook is a partition with at most two different parts , i.e. it is of the form @xmath9 , @xmath10 ) ; 3 .
@xmath11 and @xmath12 , or @xmath13 and @xmath14 ; 4 .
@xmath13 , one of the partitions is @xmath15 , and the other one is one of @xmath16 or a hook ; 5 .
one of the partitions is a rectangle , and the other one is one of @xmath17 ; 6 .
the partition pair is one of the pairs @xmath18 , @xmath19 , and @xmath20 .
we also provide the explicit combinatorial formulae for calculating any multiplicity - free kronecker product in [ sec : mf - products ] . using this
we can then easily prove the following consequence of [ thm : classification ] : [ thm:3classification ] let @xmath21 be partitions of @xmath3 , all different from @xmath7 and @xmath22 .
then the kronecker product @xmath5\cdot [ \mu]\cdot [ \nu]$ ] of the irreducible characters @xmath23,[\mu ] , [ \nu]$ ] of @xmath6 is not multiplicity - free . assuming the classification of multiplicity - free kronecker products for a symmetric group @xmath6 , with some further work the complete list of multiplicity - free products involving _ skew characters _ of @xmath6 is obtained
; we state this below . indeed
, this will be an important tool in the inductive proof of theorem [ thm : classification ] .
a proper skew diagram is one that is not the diagram of a partition up to rotation , the corresponding skew character has two distinct irreducible constituents by ( * ? ? ?
* lemma 4.4 ) ; we shall refer to such a character as a _ proper skew character_. [ thm : classification - skew ] no product of two proper skew characters is multiplicity - free .
now , let @xmath24 be a partition of @xmath25 and let @xmath26 denote a proper skew character of @xmath27 .
the product @xmath28 $ ] is multiplicity - free if and only if one of the following holds ( up to conjugation of one of the characters ) : 1 . @xmath26 is a multiplicity - free skew character , and @xmath29 $ ] is a linear character ; 2 .
@xmath30 , @xmath31 : @xmath32 , @xmath33 = [ n]+[n-1,1]$ ] ; 3 .
@xmath13 , @xmath34 , @xmath35 , @xmath36 = [ k+1,k-1]+[k , k]$ ] .
the layout of the paper is as follows . in section [ sec : background ] , we recall the results concerning kronecker and littlewood richardson coefficients which will be useful for the remainder of the paper , chief among these are dvir recursion and manivel s semigroup property .
we also explain our methodology and the intersection diagrams which will be essential in the bulk of the paper . in section
[ sec : mf - products ] , we verify that the products on our list are indeed multiplicity - free and provide formulae for decomposing these inner tensor products ; using some of these , we also show how to deduce [ thm:3classification ] from [ thm : classification ] .
sections [ sec : warm - up ] to [ sec : finale ] are dedicated to proving the converse , namely that any product @xmath5\cdot [ \mu]$ ] such that the pair @xmath37 is not on the list in [ thm : classification ] , contains multiplicities .
section [ sec : warm - up ] serves as a gentle introduction to the techniques which will be used in sections [ sec : rectangle ] , [ sec : fatty ] , and [ sec : finale ] ; here we consider tensor squares , products involving a hook , and products involving a 2-line partition . in section [ sec : if1then2 ] , we show that if [ thm : classification ] has been proven to be true for all partitions of degree less than or equal to @xmath25 , then [ thm : classification - skew ] is also true for all skew - partitions of degree less than or equal to @xmath25 .
we then begin our inductive proof of [ thm : classification ] in earnest . in sections [ sec :
rectangle ] and [ sec : fatty ] we consider products involving either a character labelled by a rectangle or fat hook partition ; such products are the most difficult to tackle using dvir recursion and the semigroup property as one is more likely to reduce to a multiplicity - free product . finally , in section [ sec : finale ] we prove that if [ thm : classification ] and thus also [ thm : classification - skew ] are true for all partitions of degree less than or equal to @xmath38 , then they also hold true for any product involving partitions of degree @xmath25 .
the hard work in earlier sections has a surprising pay - off : the large reduction from arbitrary tensor products to those involving a fat hook is much simpler than one would expect .
the main technique in the final section is to reduce to a product involving a fat hook or a rectangle and to appeal to the earlier sections .
we let @xmath27 denote the symmetric group on @xmath25 letters .
the combinatorics underlying the representation theory of the symmetric group is based on partitions .
a _ partition _ @xmath39 of @xmath25 , denoted @xmath40 , is defined to be a weakly decreasing sequence @xmath41 of non - negative integers such that the sum @xmath42 equals @xmath25 .
the _ length _ of a partition @xmath40 is the number of nonzero parts , we denote this by @xmath43 .
the _ width _ of a partition @xmath40 is the size of the first part and is denoted @xmath44 .
the _ depth _ of a partition @xmath40 is @xmath45 .
we identify a partition , @xmath39 , with its associated _ young diagram _ , that is the set of nodes @xmath46 a node @xmath47 of @xmath39 is _ removable _ if it can be removed from the diagram of @xmath39 to leave the diagram of a partition , while a node not in the diagram of @xmath39 is an _ addable _ node of @xmath39 if it can be added to the diagram of @xmath39 to give the diagram of a partition .
the set of removable ( respectively addable ) nodes of a partition , @xmath39 , is denoted by @xmath48 ( respectively @xmath49 ) . given @xmath50 ( respectively @xmath51 ) we let @xmath52 ( respectively @xmath53 ) denote the partition obtained by removing the node @xmath54 from ( respectively adding the node @xmath54 to ) the partition @xmath39 . given @xmath55 , we define the _ conjugate _ or _ transpose _ partition , @xmath56 , to be equal to the partition obtained from @xmath39 by reflecting its young diagram through the @xmath57 diagonal . the durfee length of @xmath58 is the diagonal length of the young diagram of @xmath39 , and thus gives the side lengths of the largest square which fits into the young diagram of @xmath39 . given @xmath59 and @xmath60 partitions such that @xmath61 for all @xmath62 , we write @xmath63 . if @xmath63 , then the _ skew partition _ or _ skew young diagram _ ( denoted @xmath64 ) is simply the set difference between the young diagrams of @xmath60 and @xmath59 . if @xmath65 then we say that @xmath64 is a skew partition of @xmath25 .
we let @xmath66 denote the diagram obtained by rotating the young diagram of @xmath67 through @xmath68 .
we say that a skew diagram @xmath67 is a _ proper skew diagram _ if neither @xmath67 nor @xmath66 is the diagram of a partition .
we say that a skew diagram @xmath69 is _ basic _ if it does not contain empty rows or columns , in other words @xmath70 , @xmath71 for each @xmath72 . over the complex numbers ,
the irreducible characters , @xmath5 $ ] , of @xmath6 are indexed by the partitions , @xmath73 . given a skew partition @xmath74 of @xmath25 , we have an associated _ skew character _ @xmath75 $ ] of @xmath27 , see ( * ? ? ?
* section 2.4 ) for more details .
for the corresponding definitions of schur and skew schur functions , see @xcite .
we recall the classification of multiplicity - free outer products of irreducible characters and multiplicity - free skew characters of symmetric groups as in @xcite and @xcite , respectively .
a complete list of multiplicity - free outer products of two irreducible characters of symmetric groups is given as follows : * @xmath76 \boxtimes [ \text{rectangle}]$ ] ; * @xmath76 \boxtimes [ \text{near - rectangle}]$ ] ; * @xmath77\boxtimes [ \text{fat hook}]$ ] ; * @xmath78\boxtimes [ \text{anything}]$ ] .
here , a linear partition ( 2-line rectangle ) means a partition with one row or one column ( two rows or two columns ) .
a near - rectangle is obtained from a rectangle by adding a single row or column to a rectangle , so a near - rectangle is a special fat hook .
generalising this result , gutschwager @xcite classified the basic skew partitions giving multiplicity - free skew characters ; this is closely connected to the classification of multiplicity - free products of schubert classes given by thomas and yong @xcite .
let @xmath79 be a basic skew diagram ; it may be connected or decompose into two or more pieces ( where two adjacent pieces only meet in a point ) .
we define two paths along the rim of @xmath79 .
the _ inner path _ starts in the lower left corner with an upward segment , follows the shape of @xmath80 and ends with a segment to the right in the upper right corner ; here , by a _ segment _ we mean the maximal pieces of the path where the direction does nt change .
the _ outer path _ starts in the lower left corner with a segment to the right , follows the shape of @xmath81 and ends with an upward segment in the upper right corner .
we let @xmath82 and @xmath83 denote the length of the shortest straight segment of the inner path and of the outer path , respectively .
figure [ basic skew diagrams ] depicts several basic skew diagrams , where the partition @xmath81 is shown embedded in a rectangle , with complementary partition @xmath84 . in the middle picture
, the skew diagram @xmath79 decomposes into two pieces @xmath85 and @xmath86 .
@xmath87 \draw ( 0,0)--(5,0)--(5,-3)--(0,-3)--(0,0 ) ; \draw ( 4,0)--(4,-0.5)--(3,-0.5)--(3,-1.5)--(1,-1.5)--(1,-2)--(0,-2 ) ; \draw ( 5,-2)--(4,-2)--(4,-2.5)--(3,-2.5)--(3,-3 ) ; \draw(1.25,-0.75 ) node { $ \sigma$ } ; \draw(2.25,-2 ) node { $ \rho/\sigma$ } ; \draw(4.25,-2.75 ) node { $ \tau$ } ; \end{tikzpicture } \qquad \begin{tikzpicture}[scale=0.6 ] \draw ( 0,-1.5)--(0,0)--(4,0 ) ; \draw ( 5,-1.5)--(5,-3)--(3,-3 ) ; \draw ( 5,0)--(4,0)--(4,-1.5)--(0,-1.5)--(0,-3 ) ; \draw ( 5,0)--(5,-1.5)--(4,-1.5)--(4,-2)--(3,-2)--(3,-3)--(0,-3 ) ; \draw(2.25,-0.75 ) node { $ \sigma $ } ; \draw(1.5,-2.2 ) node { $ \delta '' $ } ; \draw(4.5,-0.75 ) node { $ \delta ' $ } ; \draw(4.25,-2.5 ) node { $ \tau $ } ; \end{tikzpicture } \qquad \begin{tikzpicture}[scale=0.6 ] \draw ( 0,-1.5)--(0,0)--(4,0 ) ; \draw ( 5,-2)--(5,-3)--(3,-3 ) ; \draw ( 5,0)--(4,0)--(4,-1.5)--(0,-1.5)--(0,-3 ) ; \draw ( 5,0)--(5,-2)--(4,-2)--(4,-2.5)--(3,-2.5)--(3,-3)--(0,-3 ) ; \draw(2.25,-0.75 ) node { $ \sigma $ } ; \draw(2.25,-2.2 ) node { $ \rho/\sigma $ } ; \draw(4.25,-2.75 ) node { $ \tau $ } ; \end{tikzpicture } \end{minipage}\ ] ] [ fomentation ] before we state the classification of the basic skew diagrams labelling multiplicity - free skew characters , we recall that the character associated to a skew diagram is homogeneous if and only if the diagram is a partition diagram up to a possible rotation by @xmath88 ; in which case it is already irreducible ( see @xcite ) .
thus , the skew diagram is proper if and only if the corresponding skew character is proper , i.e. , it has at least two different constituents .
@xcite [ thm : mf - basic - skew ] let @xmath89 be a basic proper skew diagram .
then the skew character @xmath90 $ ] is multiplicity free if and only if up to rotation of @xmath89 by @xmath88 , we have @xmath91 with @xmath80 a rectangle , and additionally one of the following conditions holds : 1 .
@xmath92 ; 2 .
@xmath93 , @xmath94 ; 3 .
@xmath95 , @xmath94 ; 4 .
@xmath96 .
we emphasise that theorem [ thm : mf - basic - skew ] covers all cases of multiplicity - free proper skew characters ; in particular , the skew character @xmath97 $ ] is not multiplicity - free when the diagram @xmath79 decomposes into more than two connected components , or if it decomposes into two components and one of them is a proper skew partition .
furthermore , note that in the cases ( 2)-(4 ) described above , the complementary partition @xmath84 to @xmath79 ( in the pictures above ) is a ( rotated ) fat hook , as in [ basic skew diagrams ] .
assuming that the two pictures to the right in figure [ basic skew diagrams ] are scaled such that the short segments on the outer path are of length 1 , the theorem tells us that these skew diagrams correspond to multiplicity - free characters , whereas the skew diagram in the left picture certainly does not as both @xmath80 and @xmath84 are not rectangular .
we now recall manivel s semigroup property for kronecker coefficients @xcite
. this will be one of the two main tools used in proving the classification theorem .
let @xmath98 be partitions of @xmath25 .
we define the kronecker coefficients @xmath99 to be the coefficients in the expansion @xmath100\cdot [ \mu ] = \sum_{\nu \vdash n } g(\lambda,\mu,\nu ) \ , [ \nu]\:.\ ] ] in principle , they may be computed via the scalar product , in other words , @xmath101\cdot [ \mu ] , [ \nu]\rangle = \frac{1}{n!}\sum_{g\in { \mathfrak{s}}_n } [ { \lambda}](g ) [ \mu](g ) [ \nu](g),\ ] ] from which it also shows that the kronecker coefficients are symmetric in @xmath102 .
for @xmath103 we also define @xmath104 so that the kronecker product @xmath23\cdot[\mu]$ ] is multiplicity - free if and only if @xmath105 .
[ prop : monotonicity ] let @xmath106 and @xmath107 .
if both @xmath108 and @xmath109 then @xmath110 in particular , @xmath111 we will often use this as a reduction procedure , in particular by removing rows and columns from two partitions under consideration . as @xmath112 , we can conjugate one or both of the partitions in the result above .
this means that for the inequality , we do not have to take both partitions away from rows at the top but may take off one ( or both ) from columns at the bottom . for a given partition @xmath113 and @xmath114 ,
we let @xmath115 and @xmath116 .
[ semigroup ] let @xmath117 be partitions of @xmath25 , and suppose there exist some @xmath118 and @xmath119 such that @xmath120 .
then @xmath121 in particular , if either @xmath122 or @xmath123 , then it follows that @xmath124 also .
suppose that @xmath125 .
let @xmath126 and @xmath127 be partitions such that @xmath128 .
we have that @xmath129 and so the result follows .
if @xmath130 , we say that @xmath131 is an @xmath132-_removable _ ( or _ semigroup removable _ ) skew partition .
see [ hjflsdahjlfakds ] for an example of how one can use this procedure to prove that a product contains multiplicities .
we now recall dvir s recursive approach to calculating the value of a given kronecker coefficient .
this is the second main tool which we shall use in our proof of the classification theorem . in the following , if @xmath133 is a partition , we set @xmath134 .
[ theo : thdvir1 ] ( ( * ? ? ? * and 2.4 ) , ( * ? ? ? * and 2.1(d ) ) ) [ cdvir2][dvirmaximal ] let @xmath58 , @xmath135 be partitions of @xmath25 .
then @xmath136 let @xmath137 and set @xmath138 . if @xmath139 , then @xmath140 \cdot [ \mu / \beta ] , [ \hat{\nu}]\rangle \ ; .\ ] ] [ divr0]in the situation above , note that by the littlewood - richardson rule and the bound on the width given above , any constituent @xmath29 $ ] of @xmath141 \cdot [ \mu / \beta]$ ] has width at most @xmath142 , so @xmath143 can always be extended to a partition @xmath144 , giving a constituent in @xmath23\cdot[\mu]$ ] .
since skew characters of @xmath6 can be decomposed into irreducible characters using the littlewood - richardson rule , the following theorem provides a recursive formula for the coefficients @xmath99 .
[ theo : thdvir2 ] @xcite .
let @xmath145 and @xmath146 be partitions of @xmath25 .
define @xmath147 i.e. , @xmath148 is the set of partitions obtained from @xmath149 by adding a horizontal strip of size @xmath150 .
then @xmath151 \cdot [ \mu/\alpha ] , [ \hat{\nu}]\rangle - \sum_{{{\eta \in y(\nu ) \atop { \eta \neq \nu } } \atop \eta_1 { \leqslant}|{\lambda}\cap \mu| } } g ( { \lambda},\mu,\eta ) \ ; .\ ] ] this is crucial for the following result that will be useful later .
* lemma 4.6 ) , ( * ? ? ?
* lemma 2.6 ) [ lem : prepalmext ] let @xmath152 be partitions not of the form @xmath7 or @xmath8 up to conjugation .
set @xmath153 .
assume that @xmath154 is a single row and that @xmath155 $ ] is an irreducible character @xmath156 $ ] , with a partition @xmath157 .
then we have @xmath158 .
furthermore , we define the virtual character @xmath159 \cdot [ \mu / \beta_a ] - \sum_{b \in { \operatorname{add}}(\alpha ) } \alpha^b \:.\ ] ] then if @xmath160\rangle>0 $ ] , for @xmath161 , then @xmath162 is a partition of @xmath25 , and @xmath163\rangle$ ] .
we shall frequently use the following terms : * linear partition ( or linear character ) to mean a partition of the form @xmath164 or @xmath165 ( or the corresponding character @xmath166 $ ] or @xmath167 $ ] ) for some @xmath168 ; * the natural character to mean the character @xmath169 $ ] for some @xmath170 ; * 2-line partition to mean a partition , @xmath39 , such that @xmath171 or @xmath172 ; * proper hook to mean a partition of the form @xmath173 for @xmath174 ; * fat rectangle to mean a rectangle which is not linear or a 2-line rectangle ; * proper fat hook to mean a fat hook which is not equal to a rectangle , hook , or 2-line partition ; * proper skew partition to mean a skew partition , @xmath39 , such that neither @xmath39 nor @xmath175 is a proper partition . given @xmath176 , we shall refer to the diagram for this pair of partitions to be the diagram obtained by placing the partitions @xmath58 and @xmath135 on top of one another so that one can see the intersection of these partitions ( usually denoted @xmath177 ) and the set differences @xmath178 and @xmath179 explicitly , see for example [ 3aa3 ] .
[ hjflsdahjlfakds ] suppose we wish to show that the tensor square of the character @xmath180 $ ] contains multiplicities .
we do this by considering the possible ways in which we can reduce our problem ( using dvir recursion or the semigroup property ) to a problem for a pair of smaller partitions .
we have that @xmath181 and @xmath182 by the semigroup property , as required ( for example , the coefficient @xmath183 ) .
alternatively , one can prove that @xmath23 \cdot [ \mu]$ ] contains multiplicities ( for @xmath184 ) as follows .
if @xmath185 then the result can be verified by direct computation . for @xmath186 , we can conjugate and obtain @xmath187 by dvir recursion .
the result then follows by induction .
suppose we wish to show that the product @xmath188 \cdot [ 11,7 ^ 3,6,5 ^ 4,2,1]$ ] contains multiplicities .
the diagram is the rightmost depicted in [ 3aa3 ] .
we have that @xmath189 and so @xmath190 using dvir recursion , as required .
alternatively , one can use [ semigroup ] to remove all rows and columns which are common to both @xmath58 and @xmath135 to obtain the pair of partitions @xmath191 and @xmath192 .
the result then follows from the previous example .
@xmath193 \draw[very thick ] ( 0,0)--(1.5,0)--(1.5,-1.5)--(0,-1.5)-- ( 0,0 ) ; \draw [ ] ( 1.5,0)--(2.5,0 ) ( 1.5,-1.5)--(2.5,-1.5 ) ( 4,0)--(5.5,0)--(5.5,-1.5)--(4,-1.5 ) ; \draw(4.75,-0.75 ) node { $ \gamma$ } ; \draw[dotted ] ( 2.5,-1.5)--(4,-1.5 ) ( 2.5,0)--(4,0 ) ; \draw ( 0,-1.5)--(0,-2.5 ) ( 1.50,-1.5)--(1.50,-2.5 ) ( 0,-4)--(0,-5.5)--(1.5,-5.5)--(1.5,-4 ) ; \draw[dotted ] ( 1.5,-2.5)--(1.5,-4 ) ( 0,-2.5)--(0,-4 ) ; \draw(0.75,-4.75 ) node { $ \delta$ } ; \end{tikzpicture}\end{minipage}\quad \quad \begin{minipage}{38mm}\begin{tikzpicture}[scale=0.5 ] \draw[very thick ] ( 0,0)--(5.5,0)--(5.5,-0.5)--(3.5,-0.5 ) --(3.5,-2 ) --(3,-2 )
--(3,-2.5 ) --(2.5,-2.5)--(2.5,-3)--(1,-3)--(1,-5)--(0.5,-5)--(0.5,-5.5)--(0,-5.5)--(0,0 ) ; \draw ( 5,-0.5)--(5,-2)--(3.5,-2 ) ; \draw ( 2.5,-3)--(2.5,-4.5)--(1,-4.5 ) ; \draw(4.25,-1.25 ) node { $ \gamma$ } ; \draw(1.75,-3.75 ) node { $ \delta$ } ; \end{tikzpicture}\end{minipage}\ ] ] in [ sec : warm - up ] we shall first prove [ thm : classification ] in the case of products @xmath5\cdot [ \mu]$ ] such that @xmath194 ; one of @xmath58 or @xmath135 is a hook ; and one of @xmath58 or @xmath135 is a 2-line partition .
this allows us to avoid the discussion of small critical cases in the later sections and serves as an introduction to the methods used . in [ sec : if1then2 ] , we shall then show that if [ thm : classification ] holds by induction on the degree , @xmath25 , then so does [ thm : classification - skew ]
. we shall then begin our inductive proof of [ thm : classification ] , assuming the validity of both [ thm : classification , thm : classification - skew ] for partitions of strictly smaller degree .
we shall then consider products with a rectangle , followed by products with a fat hook , and then finally arbitrary kronecker products . at each stage
, our strategy will be to prove the result by using the semigroup property and dvir recursion to reduce the problem to @xmath195 a pair of partitions of strictly smaller degree and then appealing to our inductive proof , or @xmath196 a pair of partitions of degree @xmath25 which have already been considered .
for example in [ sec : rectangle ] we shall reduce to pairs involving a 2-line or hook partition ; in [ sec : fatty ] we shall reduce to pairs involving a rectangle , or a 2-line , or hook partition .
around the time of the classification conjecture , a number of formulae for special products and for constituents of small depth had already been obtained , notably by jeff remmel and his collaborators , as well as jan saxl and ernesto vallejo .
this allowed bessenrodt to check , prior to making the conjecture , that all the products on the list were indeed multiplicity - free . in this section
we collect together the non - trivial formulae for the products on our list ( up to conjugation ) .
some of these have appeared in the literature in the past years , and in these cases we refrain from giving proofs and provide references instead .
we start by recalling the products with the character @xmath197 $ ] , which are easy to compute , and then the classification of such multiplicity - free products is not hard to deduce ( see @xcite ) .
( * ? ? ? * lemma 4.1)[lem : lnat ] let @xmath198 , and let @xmath135 be a partition of @xmath25 .
let @xmath199 .
then @xmath200 \cdot [ n-1,1 ] = \left(\sum_{a\in { \mathrm{rem}}({\lambda } ) } \sum_{b\in { \operatorname{add}}(\mu_a ) } [ ( \mu_a)^b]\right ) - [ \mu]=(r-1)[\mu ] + \text { other constituents } \:.\ ] ] applying the formula above , the multiplicity - free products occurring below can easily be given explicitly in any concrete case .
we set @xmath201 if @xmath202 , and 0 otherwise .
similarly , we set @xmath203 if @xmath204 , and 0 otherwise .
we extend this notation to other inequalities in the obvious fashion .
[ cor : mfnat ] let @xmath198 , and let @xmath135 be a partition of @xmath25
. then 1 .
@xmath205 \cdot [ n-1,1]$ ] is multiplicity - free if and only if @xmath135 is a fat hook .
2 . @xmath205 \cdot [ n-1]\uparrow^{{\mathfrak{s}}_n}$ ] is multiplicity - free if and only if @xmath135 is a rectangle .
in particular , for @xmath206 we have that @xmath207 ^ 2 = [ n ] + [ n-1,1]+\chi_{(n>3)}[n-2,2]+ [ n-2,1 ^ 2]\:.\ ] ] the classification of kronecker squares was also verified in the course of making the classification conjecture in 1999 using the formulae stated below ( which follow as special cases from @xcite ) and work of saxl , zisser and vallejo @xcite . in the next section
we will provide a short proof that the square products in [ cor : mfnat ] and [ prop : special-2part - prod ] constitute a complete list of nontrivial multiplicity - free square products ( up to conjugation ) using the semigroup property .
[ prop : special-2part - prod]let @xmath208 . 1 . for @xmath11 , we have @xmath209 ^ 2 = \sum_{{\lambda}\vdash 2k+1\atop { \ell(\lambda ) { \leqslant}4 } } [ \lambda]\:.\ ] ] 2 .
let @xmath13 , we let @xmath210 and @xmath211 denote the sets of partitions of @xmath25 into only even parts and only odd parts , respectively , then @xmath212 ^ 2 = \sum_{{\lambda \in e(n ) } \atop { \ell(\lambda ) { \leqslant}4 } } [ \lambda ] + \sum_{{\lambda \in o(n ) } \atop { \ell(\lambda ) = 4 } } [ \lambda]\:.\ ] ] 3 .
let @xmath13 , we have that @xmath212\cdot [ k+1,k-1 ] = \sum_{{{\lambda}\vdash n , \lambda \not\in
e(n)}\atop { \ell(\lambda ) < 4 } } [ \lambda ] + \sum_{{{\lambda}\vdash n , \lambda \not\in o(n)\cup e(n)}\atop { \ell(\lambda ) = 4 } } [ \lambda ] \:.\ ] ] the decompositions @xmath195 , @xmath196 , @xmath213 have since appeared explicitly in @xcite , so we refrain from elaborating on the proof .
note that the products in @xmath196 and @xmath213 are in the following sense complementary ; we have @xmath212 \cdot ( [ k , k]+[k+1,k-1 ] ) = \sum_{{{\lambda}\vdash 2k}\atop{\ell(\lambda ) { \leqslant}4 } } [ \lambda ] \:.\ ] ] the decomposition of the products of characters involving a 2-line partition and a hook partition has been determined explicitly by remmel @xcite and rosas @xcite .
the formulae there are quite involved , but can be applied in our special case to show [ prop : kk - hook ] let @xmath13 , and let @xmath214 be a hook .
then @xmath215\cdot [ \mu]$ ] is multiplicity - free . by the formulae given in @xcite or @xcite
it is clear that no constituent to a partition of durfee length 3 can appear , but only hooks and double - hooks . from the formula in (
* theorem 2.2 ) for the multiplicity of hook constituents in the product , it is immediate that each of these can appear at most once ( note that in theorem 2.2(ii)(c ) the second term ca nt appear for @xmath216 ) . for a double - hook @xmath113 , we might use either ( * ? ? ?
* theorem 2.2 ) or ( * ? ? ?
* theorem 4 ) to deduce that @xmath217 or 1 .
let @xmath218 be our hook , and let @xmath113 be a double - hook that is not a hook , written as @xmath219 ( here @xmath220 , @xmath221 ) ; we may assume ( by conjugation if necessary ) that @xmath222 .
we recall the formula from ( * ? ? ?
* theorem 4 ) : @xmath223 where @xmath224 first we consider the case where @xmath225 and @xmath226 .
then @xmath227 , so @xmath228 , and hence @xmath229 . if @xmath230 and @xmath231 , then @xmath232 , hence @xmath233 , so we must have @xmath234 and then @xmath235 .
but then @xmath226 , a contradiction .
if @xmath236 , then we also have @xmath237 . in this case , we must have @xmath238 and @xmath239 .
by our assumption , @xmath240 , hence @xmath241 since @xmath242 , we obtain @xmath243 , and thus we have the contradiction @xmath244 hence the multiplicity @xmath245 is always at most 1 . we now provide explicit formulae for the kronecker products of small depth listed in [ thm : classification ] . also these products were checked in the course of making the classification conjecture in 1999 using @xcite .
we use this opportunity to correct a small mistake in the statement of the formula for the decomposition given in ( * ? ? ?
* corollary 4.6 ) ; this correction is provided in case @xmath195 below . [ products with a partition of small depth ] the kronecker products involving a partition of depth 2 or 3 listed in [ thm : classification ] can be calculated as follows . here
we take the convention that if @xmath58 is not a partition , then @xmath23 $ ] is zero .
* let @xmath246 , @xmath247 , with @xmath248 .
then the decomposition of the product @xmath249\cdot [ a^b]$ ] is as follows , @xmath250+\chi_{(a>2)}[a^{b-1},a-1,1 ] + [ a^{b-2},(a-1)^2,1 ^ 2 ] + \chi_{(b>3)}[(a+1)^2,a^{b-4},(a-1)^2 ] \\ &
+ \chi_{(b>2 ) } [ a+1,a^{b-2},a-1]+\chi_{(b>2)}[a+1,a^{b-3},(a-1)^2,1]+[a+2,a^{b-2},a-2 ] \\ & + \chi_{(a>2)}[a+1,a^{b-2},a-2,1]+\chi_{(a>3)}[a^{b-1},a-2,2 ] \:.\end{aligned}\ ] ] * let @xmath30 and @xmath247 , with @xmath251 .
then the decomposition of the product @xmath252\cdot [ a^b]$ ] is as follows , @xmath253+[a+1,a^{b-2},a-1]+[a+1,a^{b-2},a-2,1 ] + [ a^{b-2},(a-1)^2,2]\\ & + \chi_{(b>2)}[a+1,a^{b-3},(a-1)^2,1]+\chi_{(b>2)}[(a+1)^2,a^{b-3},a-2 ] + [ a^{b-1},a-2,1 ^ 2]+[a^{b-1},a-1,1 ] \:.\end{aligned}\ ] ] * let @xmath254 .
then the decomposition of the product @xmath255\cdot [ k , k]$ ] is as follows , @xmath256+[k+1,k-2,1]+[k , k-1,1 ] + [ k , k-2,1 ^ 2]+[k , k-2,2]+[k , k-3,3]+\\ & [ k-1,k-1,2]+[k-1,k-2,2,1 ] + [ k+3,k-3]+[k+2,k-3,1]+[k+1,k-3,2 ] \:.\end{aligned}\ ] ] for @xmath257 the remaining multiplicity - free products of the form @xmath255\cdot [ \lambda]$ ] are precisely those with @xmath258 ( up to conjugation ) , for the corresponding @xmath25 .
case @xmath195 can be proved directly using dvir s recursion and the littlewood - richardson rule .
case @xmath196 may be proved easily by applying dvir s recursion formula , computing the multiplicity of constituents @xmath23 $ ] using @xmath259 and the known formula for @xmath260 .
case @xmath213 can be proved easily using @xcite ; as it has since appeared in ( * ? ? ?
* theorem 4.8 ) we refrain from elaborating on the proof .
finally we show in this section that our main result implies that no product of three non - linear irreducible characters of the symmetric groups is multiplicity - free ; hence at the end of this article also [ thm:3classification ] is confirmed .
[ prop:3from2 ] assume that [ thm : classification ] is true .
then also [ thm:3classification ] holds .
let @xmath261 be partitions of @xmath25 , all different from @xmath7 and @xmath22 .
first we show that a product involving a square can not be multiplicity - free . since no product of two non - linear irreducible characters of @xmath6 is irreducible by @xcite or @xcite ,
we have @xmath262[{\lambda}][\mu],[\mu ] \rangle = \langle [ { \lambda}][\mu],[{\lambda}][\mu ] \rangle > 1,\ ] ] so we are done in this case .
hence @xmath261 have to be three different partitions with pairwise multiplicity - free products . to avoid discussion of small cases , for @xmath263 the assertion of [ thm:3classification ]
is checked by computer , so we assume now @xmath264 . by conjugating if necessary , we only have to consider the following triples @xmath265 : @xmath266 , @xmath267 , @xmath268 with @xmath269 , @xmath270 , @xmath271 . in all of these cases , @xmath23[\mu]$ ] is easily seen to have a constituent @xmath272 $ ] where @xmath157 is not a fat hook by [ lem : lnat ] .
it immediately follows from [ thm : classification ] that @xmath272[\nu]$ ] , and hence @xmath23[\mu][\nu]$ ] , can not be multiplicity - free .
as a warm - up to the later sections , we shall now give a self - contained proof of the classification theorem for products @xmath5 \cdot [ \mu]$ ] involving a hook or 2-line partition or for which @xmath273 .
we first consider products of the form @xmath5\cdot [ \lambda]$ ] .
we use the semigroup property to give a simple proof of the converse to [ prop : special-2part - prod ] ( that any product not on the list contains multiplicities ) . [ selftensor ] let @xmath39 be a partition of @xmath25 . then @xmath23\cdot [ { \lambda}]$ ] is multiplicity - free if and only if @xmath58 or its conjugate is one of the following @xmath274 by [ prop : special-2part - prod ] , it will suffice to show that any product not of the above form contains multiplicities .
suppose that @xmath39 is a 2-line partition not of the above form .
up to conjugation , we can assume that @xmath275 such that @xmath195 @xmath276 and @xmath196 @xmath277 .
the smallest partition satisfying these properties is @xmath278 ; this shall be our _ seed _ and we shall _ grow _ all other 2-line cases from this one . given any @xmath39 of the above form , we have that @xmath279 where the latter term on the right - hand side is a partition because of @xmath196 . by [ prop : monotonicity ] , we have that @xmath280 ( for example , @xmath281 ) and so the product @xmath5 ^ 2 $ ] is contains multiplicities .
it remains to consider the case in which @xmath39 is a partition with @xmath282 .
set @xmath283 .
then @xmath284 now @xmath285 is a partition with @xmath286 . up
to conjugacy we only need to consider @xmath287 , with @xmath288 equal to 2 , 5 , 3 , 3 , 2 , respectively .
we will later use some more detailed information on squares . by work of saxl @xcite ,
zisser @xcite , vallejo @xcite we have the following result on constituents in squares .
we refer to ( * ? ? ?
* section 2.3.17 ) for the definition of a hook in a diagram .
[ prop : squares ] let @xmath73 , @xmath289 .
let @xmath290 for @xmath291 and @xmath292 .
then @xmath293 ^ 2 & = [ n]+a_1[n-1,1 ] + a_2[n-2,2 ] + b_2[n-2,1 ^ 2]\\[5pt ] & \ : + a_3[n-3,3 ] + b_3[n-3,1 ^ 3]+c_3[n-3,2,1 ] + \text{constituents of greater depth } \end{array}\ ] ] with @xmath294 , @xmath295 , + @xmath296 , @xmath297 , + @xmath298 , @xmath299 , + , @xmath300 .
+ in particular , for @xmath301 we always have @xmath302 . [
prop : squares ] was used to verify the classification of multiplicity - free square products in [ selftensor ] prior to 1999 . applying this result to a partition @xmath58 for which @xmath23 ^ 2 $ ]
is multiplicity - free , immediately yields that @xmath58 is a rectangle or @xmath8 or @xmath303 ( or conjugate ) .
squares of rectangles can then be dealt with using dvir recursion .
we shall now cover the case of products @xmath5\cdot [ \mu]$ ] such that one of @xmath39 or @xmath135 is a hook , different from @xmath7 , @xmath8 and their conjugates , and the other is an arbitrary partition .
let @xmath304 , and let @xmath305 with @xmath306 .
let @xmath40 , @xmath58 not equal to @xmath7 or @xmath307 up to conjugation .
if @xmath205 \cdot [ \lambda]$ ] is multiplicity - free then ( up to conjugation of @xmath145 ) we have that @xmath60 is equal to @xmath15 for @xmath13 , or @xmath308 and @xmath39 is a rectangle . from our computational data , we know that the result holds for all @xmath309 , so we may assume that @xmath310
. we will proceed by induction , so we assume that the result holds for products with hooks of size smaller than @xmath25 .
furthermore , by conjugating if necessary , we may ( and will ) assume that for both @xmath305 and @xmath58 the length is at most as large as the width , so @xmath311 .
we have to show @xmath312 for any @xmath58 different from @xmath313 and their conjugates , and with @xmath314 not on the classification list above .
we start with the case in which @xmath315 , so @xmath316 where by our assumptions @xmath317 .
we remove the third column of @xmath58 , of height @xmath318 , to obtain a partition @xmath319 , and we set @xmath320 . by our assumptions , @xmath321 is a pair not on our classification list for @xmath322 , hence by [ semigroup ] we conclude @xmath323 and we are done in this case .
we now assume @xmath324 ; since @xmath325 is not on our classification list , @xmath58 is not a rectangle when @xmath308 . we remove the third row @xmath326 from @xmath58 to obtain @xmath319 . as @xmath327 and @xmath311
, we have @xmath328 hence @xmath329 still satisfies the conditions of the proposition we want to prove . hence by induction and [ semigroup ] , we have that @xmath330 unless @xmath331 for @xmath332 , or @xmath308 and @xmath319 is an arbitrary rectangle
. indeed , both cases can only occur when @xmath333 , with @xmath334 , and if @xmath308 , we also have @xmath335 ( note that @xmath336 as @xmath310 ) . in which case , we let @xmath319 be obtained by removing the second column from @xmath58 , of height @xmath337 , and we set @xmath320
. then @xmath338 , and @xmath58 is not a rectangle in case @xmath308 , so hence again we have by induction and [ semigroup ] that @xmath330 .
we now consider products in which one factor is labelled by a partition @xmath135 with two rows or two columns .
conjugating if necessary , we may assume that @xmath135 has two rows .
[ short_second_row ] let @xmath3 and @xmath39 be a partition of @xmath25 , not equal to @xmath7 or @xmath8 up to conjugation . 1 .
let @xmath339 .
if the product @xmath249 \cdot [ \lambda]$ ] is multiplicity - free , then @xmath39 is a rectangle or @xmath58 is equal to @xmath340 ( up to conjugation ) .
2 . let @xmath341 .
if the product @xmath255 \cdot [ \lambda]$ ] is multiplicity - free , then @xmath342 or @xmath343 , @xmath344 , @xmath345 , @xmath346 ( up to conjugation ) .
as the smaller cases hold by computer calculations , we may assume that @xmath347 .
in both cases of the lemma we proceed by induction . by [ sectionhook ! ] , we may assume that @xmath39 is not a hook .
conjugating if necessary , we assume that @xmath348 ; thus @xmath349 by our assumption on @xmath25 .
we have @xmath350 with @xmath351 , and we take a partition @xmath58 ( satisfying the assumptions above ) that is not a rectangle when @xmath352 , and not @xmath15 when @xmath353 .
we remove the fourth column of @xmath58 , of height @xmath354 say , to obtain @xmath319 , and we set @xmath355 . by our assumptions , @xmath319 is not a hook , nor is it a rectangle , nor @xmath340 or its conjugate when @xmath352 , and not of the form @xmath356 or one of the exceptional small partitions or one of their conjugates when @xmath353 .. hence by induction @xmath357 , and we are done by [ prop : monotonicity ] .
[ two_equal_parts ] let @xmath358 .
let @xmath40 .
if the product @xmath215 \cdot [ \lambda]$ ] is multiplicity - free , then @xmath39 is a hook , @xmath359 or @xmath360 ( up to conjugation ) .
as the smaller cases hold by computer calculations , we may assume that @xmath361 .
we set @xmath362 .
we now assume that @xmath39 is not one of the partitions listed above that are already known to give a multiplicity - free product with @xmath215 $ ]
. we shall again proceed by induction .
we assume that @xmath348 and therefore ( by our assumption on @xmath25 ) we conclude that @xmath363 .
if the fifth or sixth column is of even height ( and equal to @xmath364 say ) , remove this column from @xmath39 to obtain @xmath319 .
otherwise , both the fifth and sixth columns are of odd height ( and their sum is equal to @xmath364 say ) ; then remove both columns from @xmath39 to obtain @xmath319 in both cases set @xmath365 .
we then have a pair of partitions @xmath366 such that @xmath367 by induction ( keeping in mind that @xmath368 ) , and hence , by proposition [ prop : monotonicity ] , @xmath369 .
[ prop : res2part ] for @xmath135 a 2-line partition , a product @xmath205\cdot[\lambda]$ ] is multiplicity - free if and only if the product is on the classification list of theorem [ thm : classification ] .
we may assume that @xmath370 by our computational data , and we proceed again by induction . by [ sec : mf - products ] it is enough to prove that any product not on the list contains multiplicities . by [ sectionhook !
] we can assume that @xmath39 is not a hook .
as before , we may ( and will ) assume that @xmath348 ; note that then @xmath371 by our assumption on @xmath25 . by [ short_second_row , two_equal_parts ]
we may assume for @xmath350 that @xmath372 ; we can also assume that @xmath58 is not of the form dealt with in these lemmas .
by [ selftensor ] , we may also assume that @xmath373 .
we first suppose that @xmath171 .
for @xmath374 with @xmath375 , we define @xmath376 by @xmath377 if @xmath378 , and by @xmath379 otherwise . with this notation in place , @xmath380 ( given our assumption on @xmath25 ) .
we now assume @xmath381 . remove the fifth column of @xmath39 , of height @xmath354 say , to obtain @xmath319 .
we have three cases to consider : @xmath195 @xmath382 , @xmath196 @xmath383 , or @xmath213
@xmath384 for @xmath385 .
corresponding to these cases we write @xmath195 @xmath386 , @xmath196 @xmath387 , or @xmath213 @xmath388 .
we hence obtain a pair of partitions @xmath321 such that by induction @xmath357 ( keeping in mind that @xmath368 ) , and hence we are again done by [ prop : monotonicity ] .
it is the aim of this section to show that if theorem [ thm : classification ] holds for a fixed @xmath389 , then so does theorem [ thm : classification - skew ] . in the final proof of theorem [ thm : classification ]
( and hence also of theorem [ thm : classification - skew ] ) by induction we may thus always assume that both theorem [ thm : classification ] and [ thm : classification - skew ] hold for all symmetric groups of degree strictly less than @xmath25 .
first we require some preparatory results on how ( multiplicity - free ) skew characters decompose into simple constituents .
the following observation by gutschwager will be a very useful tool later on .
@xcite [ skew - neighbours ] any proper skew character of @xmath6 has two neighbouring constituents , i.e. , constituents @xmath23,[\mu]$ ] such that @xmath390 .
we now describe multiplicity - free proper skew characters with large maximal constituents ( in the lexicographic ordering of the partition labels ) .
[ lm : special - skew - mf ] let @xmath26 be a multiplicity - free proper skew character of @xmath6 . 1 .
if @xmath26 has a constituent @xmath391 $ ] , then @xmath392\boxtimes [ k]= \displaystyle \sum_{i=0}^k [ n - k+i , k - i]$ ] , for some @xmath393 .
2 . if @xmath26 has maximal constituent @xmath197 $ ] , then we have one of the following : * @xmath394 and @xmath395\boxtimes [ k]=\sum_{i=0}^k [ n-1-i,1+i]+\sum_{i=0}^{k-1 } [ n-2-i,1+i,1];\ ] ] * @xmath396 , @xmath397 and @xmath398=\sum_{i = m}^{b-1 } [ a - b+1+i , b - i];\ ] ] 3 .
if @xmath26 has maximal constituent @xmath249 $ ] , and also has @xmath252 $ ] appearing as a constituent , then we have one of the following : * @xmath399\boxtimes[1 ^ 2]=[n-2,2]+[n-2,1 ^ 2]+[n-3,1 ^ 3]$ ] ; * @xmath400=[((n-2)^2,s)/(n-3,s-1)]$],@xmath401 ; * @xmath402 $ ] , @xmath403 .
4 . if @xmath26 has maximal constituent @xmath249 $ ] , and contains also @xmath255 $ ] , but not @xmath252 $ ] , then we have one of : * @xmath404\boxtimes [ k]=[n-2,2]+[n-3,3]+ ... + [ n-3,2,1]+ ... +[n-4,2 ^ 2]$ ] ; * @xmath405=[n-2,2]+[n-3,3]+ ... = \displaystyle\sum_{a=2}^{\lfloor n/2\rfloor } [ n - a , a]$ ] .
first we note that the diagram of the proper skew character @xmath406 $ ] can have at most two components ( as no outer product of three characters is multiplicity - free ) by [ subsec : mf - skew ] .
our tactic for the proof will be to examine the maximal constituents of skew characters using the littlewood
richardson rule @xcite and then considering which characters also satisfy the conditions of [ subsec : mf - skew ] .
we shall write @xmath407 $ ] for the duration of the proof .
\(1 ) assuming that @xmath26 contains @xmath391 $ ] , we immediately deduce ( from the littlewood
richardson rule ) that the skew diagram @xmath408 consists solely of disconnected single rows . by [ thm : mf - outer ]
, there are at most two such disconnected rows in @xmath26 .
hence the skew character is of the form @xmath409\boxtimes [ k ] = \sum_{i=0}^k [ n - k+i , k - i]$ ] , for some @xmath410 .
\(2 ) assume now that the maximal constituent of the skew character @xmath26 is @xmath197 $ ] .
again by the littlewood
richardson rule , we deduce that the skew diagram @xmath408 then must have one column of length 2 , and all others are of length 1 . as the skew diagram has at most two components , this leaves only a few possibilities for the skew character , and we can only have the two types listed in ( 2 ) , by [ subsec : mf - skew ] . to ( 3 ) and ( 4 ) .
we assume that @xmath26 has maximal constituent @xmath249 $ ] .
note that the skew diagram @xmath408 then must have two columns of length 2 , and all others are of length 1 , and as before , it has at most two components ; if it is disconnected , then both components are of partition shape ( up to rotation ) .
we first consider case @xmath411 , in which @xmath26 contains @xmath252 $ ] as a constituent .
then , if the diagram has two components , both components have a column of length 2 ( by the littlewood
richardson rule ) . by theorem
[ thm : mf - outer ] , the only possibility is then that @xmath399\boxtimes[1 ^ 2]=[n-2,2]+[n-2,1 ^ 2]+[n-3,1 ^ 3]$ ] .
now assume that @xmath408 is connected and @xmath252 $ ] appears as a constituent of @xmath26 .
in which case the diagram of @xmath412 does not appear as a subdiagram of @xmath69 ( by the littlewood
richardson rule ) .
now , the multiplicity - free condition leaves only the possibilities @xmath413 , with @xmath414 , or @xmath415 , with @xmath416 , and for @xmath417 the corresponding skew characters are equal since the diagrams only differ by a rotation .
since @xmath418 , we have @xmath419 . for case ( 4 ) , we now assume that @xmath26 has maximal constituent @xmath249 $ ] and contains @xmath255 $ ] , but not @xmath252 $ ] . if the diagram is disconnected , the only possibility for @xmath26 is @xmath420\boxtimes [ k]=[n-2,2]+[n-3,3]+ ... + [ n-3,2,1 ] + \dots + [ n-4,2,2].\ ] ] when @xmath69 is connected , the diagram of @xmath412 appears as a subdiagram of that of @xmath69 , by our assumption that @xmath26 does not contain @xmath252 $ ] .
therefore @xmath421=[n-2,2]+[n-3,3]+ ... = \sum_{a=2}^{\lfloor n/2\rfloor } [ n - a , a],\ ] ] and @xmath418 implies @xmath422 . a first contribution towards classifying multiplicity - free products of skew characters with irreducible characters is contained in the following easy result .
[ mftimesfirsthook ] let @xmath26 be a proper skew character of @xmath6 .
then @xmath423 $ ] is multiplicity - free if and only if @xmath424 , and the product is then @xmath425+[1 ^ 2])\cdot [ 1 ^ 2]=[1 ^ 2]+[2]$ ] . by [ skew - neighbours ] , @xmath26 has two neighbouring constituents , which we may write as @xmath426 $ ] and @xmath427 $ ] , for @xmath157 a partition of @xmath38 and @xmath428 addable nodes for @xmath157 .
if one of the two partitions @xmath429 is not a rectangle , say @xmath430 , then @xmath431+[{\alpha}^y])\cdot [ n-1,1]$ ] contains @xmath432 $ ] , and hence the product is not multiplicity - free .
but if both @xmath430 and @xmath433 are rectangles , then we have @xmath434 and @xmath430 , @xmath433 are the rectangles @xmath435 , @xmath436 . for later usage , we now consider products of irreducible characters with characters that will appear as subcharacters in certain skew characters .
[ lm : special - rect - prod ] let @xmath437 and let @xmath157 be a partition of @xmath25 . 1 .
let @xmath438+[n-2,1 ^ 2]$ ] , and let @xmath157 be a rectangle .
then @xmath439 $ ] is multiplicity - free if and only if @xmath157 is linear , or @xmath440 and @xmath441 . in the latter case ,
@xmath442=[4]+[3,1]+[2 ^ 2]+[2,1 ^ 2]+[1 ^ 4]$ ] .
2 . let @xmath443+[n-2,2]$ ] . then @xmath439 $ ] is multiplicity - free if and only if @xmath157 is linear
. if @xmath157 is a linear partition , both products @xmath444 $ ] are clearly multiplicity - free .
\(1 ) the assertion is easily checked for @xmath441 .
summing the formulas in @xmath195 an @xmath196 of [ products with a partition of small depth ] we immediately see that @xmath445 $ ] appears with multiplicity 2 in @xmath272 \cdot \chi$ ] .
\(2 ) if @xmath157 is not a linear partition , then @xmath272 ^ 2 $ ] contains @xmath391+[n-2,2]$ ] , by proposition [ prop : squares ] .
thus @xmath446,[{\alpha}]\rangle = \langle [ n]+[n-2,2 ] , [ { \alpha}]\cdot[{\alpha}]\rangle { \geqslant}2 \ : , $ ] and hence @xmath439 $ ] is not multiplicity - free . a crucial step towards theorem [ thm : classification - skew ] , and thus a contribution towards the induction strategy mentioned previously ,
is contained in the next proposition .
[ prop : step1-to - skew ] assume that theorem [ thm : classification ] holds for a fixed @xmath389 .
let @xmath26 be a proper skew character of @xmath6 , and let @xmath272 $ ] be an irreducible character of @xmath6 .
then @xmath439 $ ] is multiplicity - free if and only if @xmath26 and @xmath157 are one of the following pairs ( up to multiplication of @xmath26 by a linear character or conjugation of the partition @xmath157 ) : 1 .
@xmath26 is any multiplicity - free proper skew character , and @xmath157 is a linear partition ; 2 .
@xmath30 for @xmath31 , @xmath447 , @xmath448\boxtimes[1]=[n]+[n-1,1]$ ] ; 3 .
@xmath13 for @xmath34 , @xmath449 , @xmath450=[k+1,k-1]+[k , k]$ ] .
we know that the products in the three situations above are indeed multiplicity - free ; for cases ( 2 ) and ( 3 ) this was already covered in section [ sec : mf - products ] ( and ( 1 ) is obvious ) .
so we now assume that @xmath439 $ ] is multiplicity - free and that @xmath272 $ ] is not linear ( in other words @xmath157 is not a linear partition ) and hence we may assume @xmath206 .
by lemma [ mftimesfirsthook ] we already know that @xmath197\cdot \chi$ ] is not multiplicity - free .
therefore we need only consider @xmath451 ( or its conjugate ) and hence we may assume that @xmath437 .
we have assumed that the classification list in theorem [ thm : classification ] is complete for our fixed @xmath3 and that @xmath28 $ ] is multiplicity - free .
every partition on the list is a fat hook and so we deduce that all constituents of @xmath26 are labelled by fat hooks . also , since @xmath26 has a non - linear constituent , @xmath157 must be a fat hook .
thus @xmath157 is a fat hook different from @xmath7 , @xmath8 ( and their conjugates , by our assumption and [ mftimesfirsthook ] respectively ) that has a multiplicity - free product with two neighbouring fat hooks ( because of [ skew - neighbours ] ) .
we shall now consider the possible partitions @xmath157 from the list in theorem [ thm : classification ] satisfying these conditions .
case - by - case , we consider @xmath24 on the list @xmath452 , @xmath453 , @xmath454 , @xmath303 , @xmath15 , @xmath455 , hooks and rectangles , and the few cases for @xmath456 . for @xmath457 , we consider the possible constituents in @xmath26 .
when @xmath458 , the only non - trivial possible constituents of @xmath26 such that @xmath459 $ ] is multiplicity - free are @xmath8 ( and its conjugate ) or a rectangle . since @xmath26 has to have two neighbouring constituents , it must contain @xmath460+[n-1,1]$ ] ( up to conjugating ) ; but @xmath249\cdot \chi_0 $ ] is not multiplicity - free ( for @xmath461 ) . for @xmath462
, the character @xmath463 $ ] has a multiplicity - free product with all @xmath464 $ ] , @xmath465 , @xmath466 .
given the previous arguments , we only have to consider the products with the neighbour pair sums @xmath467+[3,2]$ ] and @xmath463+[2 ^ 2,1]$ ] , and neither of these products are multiplicity - free . for @xmath440 , the proper skew partitions ( up to conjugation ) which give multiplicity - free products with @xmath24 are @xmath468\cdot([3 ] \boxtimes [ 1])&= [ 3,1]+[2,2]+[2,1 ^ 2 ] \\
[ 2,2]\cdot [ ( 3,2)/(1 ) ] & = [ 4]+[3,1]+[2,2]+[2,1 ^ 2]+[1 ^ 4 ] \end{aligned}\ ] ] and correspond to cases @xmath435 and @xmath411 . the remaining proper skew characters @xmath469 \boxtimes [ 1]$ ] and @xmath470 \boxtimes [ 2]$ ] which are equal to @xmath471 + [ 2,1 ^ 2]$ ] and @xmath471 + [ 4]$ ] respectively and so their products with @xmath472 $ ] are not multiplicity - free .
now we consider @xmath473 , and look again for the possible constituents in @xmath26 that have a multiplicity - free product with @xmath272 $ ] .
for @xmath474 the only possible constituents of @xmath26 whose product with @xmath29 $ ] is multiplicity - free are @xmath7 , @xmath8 , or @xmath15 and their conjugates ; with the exceptions of @xmath345 for @xmath475 ( and their conjugates ) .
recall @xmath26 has a neighbouring pair of constituents , and therefore must contain @xmath460+[n-1,1]$ ] for all @xmath476 up to conjugation .
however , @xmath255\cdot \chi_0 $ ] is not multiplicity - free for @xmath474 . for @xmath477 , the character @xmath478 $ ] has multiplicity - free products with @xmath479 $ ] , @xmath480 $ ] , @xmath481 $ ] , @xmath482 $ ] and @xmath483 $ ] ( and their conjugates ) .
the only neighbour pair sums that have multiplicity - free products with @xmath272 $ ] are @xmath479+[5,1]=[5]\boxtimes [ 1]$ ] and @xmath483+[4,2]=[(4,3)/(1)]$ ] ( up to conjugation ) which correspond to cases ( 2 ) and ( 3 ) of the proposition .
the first ( respectively second ) skew character can only be extended to @xmath479+[5,1]+[1 ^ 6]$ ] ( respectively can not be extended ) so that the product with @xmath478 $ ] remains multiplicity - free .
the former is not a skew character and so does not provide a counter example . hence , these considerations for @xmath473 have only led to the cases for @xmath477 in ( 2 ) and ( 3 ) . for @xmath484 , we consider the possible constituents in @xmath26 .
the only possible constituents in @xmath26 are then @xmath7 , @xmath8 ( and their conjugates ) and rectangles . as
before , @xmath26 must then contain @xmath460+[n-1,1]$ ] ( up to conjugating ) , except when @xmath440 , where there are further possible neighbour pairs . but @xmath252\cdot \chi_0 $ ] is not multiplicity - free , and for @xmath440 , no neighbour pair sum has a multiplicity - free product with @xmath485 $ ] .
for @xmath157 a hook partition not equal to @xmath7 , @xmath8 , @xmath454 up to conjugation ( which have already been considered ) we consider the possible constituents in @xmath26 .
the only possible constituents of @xmath26 are @xmath7 , @xmath8 or @xmath15 ( and their conjugates ) .
again , @xmath26 must then contain @xmath460+[n-1,1]$ ] ( up to conjugating ) , but @xmath272\cdot \chi_0 $ ] is not multiplicity - free . for @xmath486 for @xmath487 , we consider the possible constituents in @xmath26
. then @xmath26 could only have @xmath7 , @xmath8 or @xmath303 ( and their conjugates ) as constituents , except for @xmath488 , when also @xmath346 can also appear .
as before , @xmath26 must then contain @xmath460+[n-1,1]$ ] ( up to conjugating ) , but @xmath489\cdot \chi_0 $ ] is not multiplicity - free . for @xmath490 for @xmath491 , we consider the possible constituents in @xmath26
. then @xmath26 could only have @xmath7 , @xmath8 or @xmath15 ( and their conjugates ) as constituents .
as before , @xmath26 must then contain @xmath460+[n-1,1]$ ] ( up to conjugating ) , but @xmath492\cdot \chi_0 $ ] is not multiplicity - free .
finally we turn to rectangles .
first , let @xmath447 with @xmath493 , and assume @xmath494 .
then @xmath26 could only have @xmath7 , @xmath8 , @xmath452 or @xmath454 and their conjugates as constituents , except @xmath195 for @xmath495 when @xmath488 where @xmath496 and @xmath497 or their conjugates possibly appear , or @xmath196 @xmath498 when @xmath499 , where @xmath500 or their conjugates possibly appear .
we first exclude the cases @xmath195 and @xmath196 .
if the maximal constituent in @xmath26 is @xmath391 $ ] , by lemma [ lm : special - skew - mf ] and lemma [ lm : special - rect - prod ] we must have @xmath448\boxtimes[1]=[n]+[n-1,1]$ ] . in this case ,
@xmath501 $ ] is indeed multiplicity - free , and we are in situation ( 2 ) of the proposition .
if the maximal constituent in @xmath26 is @xmath197 $ ] , by lemma [ lm : special - skew - mf ] we only have to discuss the cases when @xmath26 is one of the skew characters @xmath502\boxtimes[1]=[n-1,1]+[n-2,2]+[n-2,1 ^ 2]$ ] , @xmath503\boxtimes[n-2]=[n-1,1]+[n-2,1 ^ 2]$ ] , or @xmath504=[n-1,1]+[n-2,2]$ ] . by lemma
[ lm : special - rect - prod](1 ) we already know that in the first case the product @xmath501 $ ] is not multiplicity - free .
we now consider the second case , where @xmath505+[n-2,1 ^ 2]$ ] . in the computation of the following scalar product we use
the information on special constituents in squares given by proposition [ prop : squares ] several times ( here , we just write @xmath506 for @xmath507 ) . @xmath508{\geqslant}&\langle
[ { \alpha}]\cdot [ n-1,1 ] , [ { \alpha}]\cdot [ n-2,1 ^ 2 ] \rangle \\ = & \langle [ { \alpha}]\cdot [ n-1]{\uparrow } - [ { \alpha } ] , [ { \alpha}]\cdot [ n-2,1 ^ 2 ] \rangle \\ = & \langle [ { \alpha}]^2 , [ n-1]{\uparrow } \cdot [ n-2,1 ^ 2 ] \rangle - \langle [ { \alpha}]^2 , [ n-2,1 ^ 2 ] \rangle \\ = & \langle [ { \alpha}]^2 , [ n-1]{\uparrow } \cdot [ n-2,1 ^ 2 ] \rangle \\ = & \langle [ { \alpha}]^2 , ( [ n-3,1 ^ 2]+[n-2,1]){\uparrow } \rangle \\ = & \langle [ { \alpha}]^2 , 2[n-2,1 ^ 2]+[n-3,2,1]+[n-3,1 ^ 3]+[n-1,1]+[n-2,2 ] \rangle \\ = & \langle [ { \alpha}]^2 , [ n-3,1 ^ 3]+[n-2,2 ] \rangle = 2,\end{aligned}\ ] ] and hence @xmath439 $ ] is not multiplicity - free in this case . in the third case ,
where @xmath505+[n-2,2]$ ] , we follow the same strategy as above and compute @xmath508{\geqslant}&\langle [ { \alpha}]\cdot [ n-1,1 ] , [ { \alpha}]\cdot [ n-2,2 ] \rangle\\ = & \langle [ { \alpha}]\cdot [ n-1]{\uparrow } - [ { \alpha } ] , [ { \alpha}]\cdot [ n-2,2 ] \rangle \\ = & \langle [ { \alpha}]^2 , [ n-1]{\uparrow } \cdot [ n-2,2 ] \rangle - \langle [ { \alpha}]^2 , [ n-2,2 ] \rangle \\ = & \langle [ { \alpha}]^2 , ( [ n-3,2]+[n-2,1]){\uparrow } \rangle - 1 \\ = & \langle [ { \alpha}]^2 , 2[n-2,2]+[n-3,3]+[n-3,2,1]+[n-1,1]+[n-2,1 ^ 2 ] \rangle -1 \\ = & \langle [ { \alpha}]^2 , 2[n-2,2]+[n-3,3 ] \rangle - 1 = 2 \:.\end{aligned}\ ] ] again , it follows that @xmath444 $ ] is not multiplicity - free .
now we may assume that @xmath26 contains none of @xmath391 , [ n-1,1]$ ] ( or their conjugates ) .
note that @xmath26 must contain a neighbour pair sum , so ( up to conjugating ) we may now assume that @xmath26 contains @xmath249+[n-2,1 ^ 2]$ ]
. given our assumption that @xmath509 , it follows from lemma [ lm : special - rect - prod ] that @xmath501 $ ] is not multiplicity - free .
we now consider the special cases @xmath195 @xmath495 and @xmath496 and @xmath497 or their conjugates appear in @xmath26 , or @xmath196 @xmath498 and @xmath500 or its conjugate appears in @xmath26 .
we remark that these cases can also be checked by computer .
we first consider case @xmath195 .
we assume that @xmath26 has none of the pair sums discussed above ; in which case @xmath26 must have one of the pair sums @xmath510+[6,3]$ ] or @xmath511+[7,2]$ ] ( up to conjugation ) . from the formula in theorem [ thm : classification ]
we immediately see that the first pair sum does not give a multiplicity - free product . by proposition [ prop :
squares ] we see that @xmath512\cdot [ 3 ^ 3]$ ] contains both @xmath511 $ ] and @xmath513 $ ] ( with multiplicity 1 ) , so @xmath512\cdot([6,3]+[7,2])$ ] contains @xmath512 $ ] with multiplicity 2 .
we now consider case @xmath196 .
we assume that @xmath26 has none of the pair sums discussed above ; in which case @xmath26 must have @xmath514 $ ] as a constituent ( up to conjugation ) .
but @xmath26 can not contain a neighbour of this constituent , so @xmath26 must contain one of the pair sums considered above , and so we are done .
this finishes the case of rectangles @xmath515 such that @xmath516 .
now let @xmath449 for @xmath487 .
then the constituents in @xmath26 could only be labelled by @xmath7 , @xmath8 , @xmath15 , @xmath455 , @xmath452 , @xmath453 @xmath454 ( and their conjugates ) and hooks , except for @xmath499 and @xmath517 , when also @xmath360 or its conjugate can appear .
we follow a similar strategy as before .
we assume first that @xmath518 .
if the maximal constituent in @xmath26 is @xmath391 $ ] , then lemma [ lm : special - skew - mf ] and lemma [ lm : special - rect - prod](2 ) imply that @xmath448\boxtimes[1]=[n]+[n-1,1]$ ] . in this case ,
@xmath501 $ ] is indeed multiplicity - free , and we are in situation ( 2 ) of the proposition . if the maximal constituent in @xmath26 is @xmath197 $ ] , lemma [ lm : special - skew - mf ] now implies that @xmath26 is one of the following three skew characters : @xmath503\boxtimes[n-2]=[n-1,1]+[n-2,1 ^ 2]$ ] ; @xmath504=[n-1,1]+[n-2,2]$ ] ; or @xmath519=[n-1,1]+[n-2,2]+[n-3,3]$ ] . for each of the first two skew characters , the simple character @xmath492 $ ] appears with multiplicity 2 in @xmath28 $ ] using [ products with a partition of small depth]@xmath195 and [ lem : lnat ]
the third character contains the second and so the product @xmath459 $ ] also contains multiplicities .
hence we may now assume that @xmath26 contains none of @xmath391 , [ n-1,1]$ ] ( or their conjugates ) .
as @xmath26 must contain a neighbour pair sum , we may now assume that @xmath26 contains ( up to conjugating ) one of the skew characters @xmath195 @xmath249+[n-2,1 ^ 2]$ ] , @xmath196 @xmath249+[n-3,3]$ ] , @xmath213 a sum of two neighbour hooks ( not involving @xmath391 , [ n-1,1]$ ] and their conjugates ) , or @xmath520 @xmath215+[k+1,k-1]$ ] . in case @xmath195 , we know by lemma [ lm : special - rect - prod ] that @xmath501 $ ] is not multiplicity - free . in case @xmath196 the simple character @xmath492 $ ] appears with multiplicity 2 in @xmath28 $ ] using [ products with a partition of small depth]@xmath195 and @xmath213 . in case @xmath213 , the character @xmath26 contains a sum of two neighbouring hooks , say @xmath521+[n - a-1,1^{a+1}]$ ] , with @xmath522 .
by [ anfmdsanfsd ] , the character @xmath523 $ ] appears with multiplicity 1 in both @xmath215\cdot [ n - a,1^a]$ ] and @xmath215\cdot [ n - a-1,1^{a+1}]$ ] . hence @xmath524 $ ]
is not multiplicity - free .
finally , we consider the last possible neighbouring pair ( from case @xmath520 ) which can appear in @xmath26 . if @xmath525=[k , k]+[k+1,k-1]$ ] , then by proposition [ prop : special-2part - prod ] the product @xmath526+[k+1,k-1 ] ) \cdot [ { \alpha}]= ( [ k , k]+[k+1,k-1 ] ) \cdot [ k , k ] = \sum_{\ell({\lambda}){\leqslant}4 } [ { \lambda}]\ ] ] is indeed multiplicity - free .
now assume the containment @xmath527+[k+1,k-1 ] \subseteq \chi$ ] is strict ; in which case @xmath26 must contain ( in addition to @xmath215+[k+1,k-1]$ ] ) one of the other possible constituents @xmath249 $ ] , @xmath252 $ ] , @xmath255 $ ] or their conjugates , or a hook
. the product of @xmath215 $ ] with any of @xmath249 $ ] , @xmath252 $ ] , @xmath255 $ ] has a constituent of length 4 , and therefore @xmath26 can not contain any of these . next we want to show that @xmath26 can not contain any of the conjugates of @xmath249 $ ] , @xmath252 $ ] , @xmath255 $ ] ; note that the first is a neighbour of the other two , so it can not occur together with one of those .
first assume that @xmath528+[k , k]+[2 ^ 2,1^{n-4}]$ ] .
note that this implies that @xmath408 has @xmath529 columns of length 2 and two of length 1 , and it has two rows of length 2 and @xmath530 of length 1 .
but since @xmath408 is the diagram of a multiplicity - free skew character , it is connected or has two components of shape as described in theorem [ thm : mf - outer ] , up to rotation of the pieces , and this is clearly impossible ( recall that @xmath531 ) . next assume that @xmath528+[k , k]+[2 ^ 3,1^{n-6}]$ ] or @xmath528+[k , k]+[3,1^{n-3}]$ ] .
then , similarly as above , we obtain a contradiction . it remains to exclude the case of an additional hook appearing in @xmath26 .
as before , we may assume that @xmath26 does not contain @xmath391 $ ] or @xmath197 $ ] ( or their conjugates ) , or pair sums already dealt with .
so assume @xmath26 has a hook constituent @xmath532 $ ] , with @xmath533 ; if there is more than one hook constituent we consider the one with minimal @xmath142 . if @xmath534 , then the hook constituent would be maximal , and hence then @xmath408 has one column of length @xmath535 , and all others are of length 1 .
but then it is clearly impossible that @xmath26 contains @xmath215 $ ] . on the other hand
, if @xmath536 , and @xmath26 also contains any of the conjugates of @xmath249 $ ] , @xmath252 $ ] , @xmath255 $ ] , the previous arguments give again a contradiction , and finally , the case @xmath528+[k , k]+[n - m,1^m]$ ] can be handled similarly as above .
now as the last case for @xmath449 , it only remains to consider the small cases @xmath537 . here , the arguments used above supplemented by computer calculations give the claim .
note that for @xmath538 , @xmath539 we may also have the possible neighbour pair sum @xmath540+[7,3]=[(7,4)/(1)]$ ] in @xmath26 , but @xmath541 $ ] is not multiplicity - free .
this concludes the case in which @xmath24 is a rectangle .
finally , it remains to consider the case where @xmath157 is a fat hook that is not of one of the special types discussed so far . excluding the cases considered so far
, we may conclude that @xmath542 and @xmath543 . therefore @xmath26 must contain @xmath460+[n-1,1]$ ] ( up to conjugation ) , but @xmath272\cdot \chi_0 $ ] is not multiplicity - free , as required .
[ prop : step2-to - skew ] assume that theorem [ thm : classification ] holds for a fixed @xmath389 .
then no product of two proper skew characters of @xmath6 is multiplicity - free . under the assumption of our proposition
, we have already classified in proposition [ prop : step1-to - skew ] the multiplicity - free products of a proper skew character and an irreducible character .
let @xmath26 be a multiplicity - free proper skew character of @xmath6 ( and therefore @xmath206 ) .
now by proposition [ prop : step1-to - skew ] , if @xmath544 is such that @xmath444 $ ] is multiplicity - free , then @xmath24 is a rectangle .
if @xmath545 is a neighbour of @xmath157 , then @xmath546 is not a rectangle ( as @xmath206 ) and so @xmath547 $ ] is not multiplicity - free .
but every proper skew character @xmath548 has two neighbouring constituents , by lemma [ skew - neighbours ] , hence @xmath549 can not be multiplicity - free .
[ cor : step - to - skew ] if theorem [ thm : classification ] holds for a fixed @xmath389 , then theorem [ thm : classification - skew ] also holds for @xmath25 .
[ induction ] for the remainder of the paper , we shall assume that [ thm : classification ] ( and hence also [ thm : classification - skew ] ) has been proven by induction for all pairs of partitions of degree strictly less than @xmath3 .
we refer to any pair @xmath550 of partitions of degree strictly less than @xmath25 and satisfying @xmath551 as a seed ( for multiplicity ) .
[ theo : thdvir1 ] implies that a necessary condition for @xmath105 is that the pair @xmath552 $ ] , @xmath553 $ ] belongs to the lists in [ thm : classification , thm : classification - skew ] .
in this section , we shall assume that @xmath554 is a partition of @xmath30 with @xmath555 .
let @xmath73 .
the product @xmath205\cdot [ \lambda]$ ] is multiplicity - free if and only if @xmath39 is one of @xmath452 , @xmath454 , @xmath8 , @xmath7 , or one of the special partitions @xmath556 ( or conjugate to one of the listed partitions ) .
one half of the proposition follows from [ sec : mf - products ] . in this section , we prove the other half of this proposition via a series of lemmas
. * important standing assumption : * for the remainder of this section , we assume that @xmath39 is not one of the listed partitions giving a multiplicity - free product , and we want to deduce that @xmath5 \cdot [ \mu]$ ] contains multiplicities .
we may assume that @xmath39 is neither a hook , or 2-line partition , and that @xmath557 , as we have already dealt with these cases in [ sec : warm - up ] .
there are two possible intersection diagrams for @xmath39 and @xmath135 , up to conjugation ; these are given in figure [ intersectiondiagramrect ] .
as indicated in the intersection diagram , we may assume ( by conjugating if necessary ) that @xmath558 for the remainder of this section .
we will also use the notation indicated there , in other words we let @xmath559 , @xmath560 ( in the second case @xmath561 ) and @xmath562 .
@xmath563 \draw[very thick ] ( 0,1)--(4.5,1)--(4.5,-3.5)--(0,-3.5)--(0,1 ) ; \draw ( 4.5,1)--(6,1)--(6,-0)--(5.5,-0)--(5.5,-1)--(5.5,-1)--(5,-1)--(5,-1.5)--(4.5,-1.5 ) ; \draw ( 4.5,-2)--(4,-2)--(4,-2.5)--(2.5,-2.5)--(2.5,-3)--(1,-3)--(1,-3.5 ) ; \draw(3.5,-3 ) node { $ \gamma$ } ; \draw(2,-1 ) node { $ \beta$ } ; \draw(5,-0 ) node { $ \delta$ } ; \end{tikzpicture}\end{minipage } \begin{minipage}{38mm}\scalefont{0.8}\begin{tikzpicture}[scale=0.55 ] \draw[very thick ] ( 0,0)--(4.5,0)--(4.5,-3.5)--(0,-3.5)--(0,0 ) ; \draw ( 4.5,0)--(7,0)--(7,-0.5)--(5.5,-0.5)--(5.5,-1.5)--(4.5,-1.5 ) ; \draw(4.5,-2)--(2,-2)--(2,-3.5 ) ; \draw ( 1.5,-3.5)--(1.5,-4.5)--(0,-4.5)--(0,-3.5 ) ; \draw(3.25,-2.75 ) node { $ \gamma$ } ; \draw(5,-0.75 ) node { $ \delta'$ } ; \draw(0.75,-4 ) node { $ \delta''$ } ; \draw(1.25,-1.25 ) node { $ \beta$ } ; \end{tikzpicture}\end{minipage}\ ] ] [ rectan ] if @xmath564 is a rectangular partition of @xmath565 for @xmath566 , then @xmath124 .
we may assume that @xmath567 .
without loss of generality we may assume that @xmath568 , and thus @xmath569 ( as we have assumed @xmath570 . as we have already dealt with 2-line partitions
, we may also assume that @xmath571 . under these assumptions ,
@xmath572 , and @xmath573 and @xmath574 are @xmath132-removable .
it then follows that @xmath575 by lemma [ selftensor ] and proposition [ prop : monotonicity ] .
[ lingamrec ] if the partition @xmath576 is @xmath165 for @xmath168 , or @xmath577 for @xmath170 , and @xmath578 has one connected component , then @xmath579 .
the structure of this proof ( and the idea behind many future proofs ) is as follows .
our assumption on @xmath67 implies that @xmath580 .
generically , we can proceed by removing the first @xmath581 columns common to both partitions @xmath58 and @xmath135 to obtain partitions @xmath319 and @xmath582 , such that @xmath582 is a 2-column partition and @xmath583 by the semigroup property . as @xmath582 is a 2-line partition , we can then ( in most cases ) apply [ prop : res2part ] to deduce that @xmath357 .
however , if @xmath584 , we shall see that this argument can fail because it is possible that we have reduced to a pair @xmath321 for which @xmath585 .
we therefore refer to the case in which @xmath584 as an ` exceptional case ' and provide a separate argument .
we begin with the generic case .
given @xmath586 such that @xmath587 , we may remove the first @xmath588 columns from @xmath39 and @xmath135 and hence obtain partitions @xmath589 and @xmath590 ( respectively @xmath591 ) for @xmath592 ( respectively @xmath593 ) .
the result then follows from the case for 2-line partitions .
now assume @xmath594 ; we have that @xmath39 is equal to either @xmath595 or @xmath596 for @xmath66 being @xmath597 or @xmath598 , respectively .
we first deal with the case @xmath599 .
we set @xmath600 and @xmath601 and rewrite our partitions as follows @xmath602 and by [ prop : monotonicity ] , we have that @xmath603 .
now , by [ prop : res2part ] we have that @xmath604 , and so the result follows .
we now deal with the case @xmath605 .
we set @xmath606 and @xmath607 and rewrite our partitions as follows @xmath608 for @xmath609 or @xmath610 , a direct computation shows @xmath611 . when @xmath612 , we have that @xmath613= [ 2^{b-3}]\boxtimes [ 2 ] \ ; , \ ; [ \tilde{\mu}^t / ( \tilde{\lambda}\cap \tilde{\mu}^t)]= [ ( b-2)^2]\:.\ ] ] the product of these characters is not multiplicity - free by proposition [ prop : res2part ] , and our inductive proof . therefore , by proposition [ prop : monotonicity ] and corollary [ dvirmaximal ] , we have @xmath614 . [ lingamrec2 ]
if @xmath576 is @xmath615 or @xmath616 , and @xmath578 has one connected component , then @xmath312 . by [ lingamrec ]
, we may assume that @xmath617 or @xmath618 .
we first consider the case @xmath619 for some @xmath620 .
we first deal with the exceptional cases which occur for small values of @xmath621 ; namely @xmath622 , and @xmath623 for @xmath624 .
we consider the exceptional cases for @xmath625 in detail . here
@xmath578 is equal to @xmath435 or @xmath436 .
if we remove all rows and columns common to @xmath58 and @xmath135 , we obtain @xmath626 . unfortunately , @xmath585 in these cases , and so we have gone too far . in other words
, we have removed too many rows or columns . if @xmath627 , then there are three ways in which we may have removed too many rows or columns , @xmath628 however , since our original partition @xmath135 contained @xmath346 ( by assumption ) , we can choose to reduce only to @xmath629 .
one can deal with @xmath630 in a similar fashion , and here reduce to the exceptional case @xmath631 .
for all these pairs we have @xmath632 by direct computation and the result follows by proposition [ prop : monotonicity ] . for @xmath633 , we remove almost all rows and columns common to @xmath58 and @xmath135 until we reach one of the following pairs @xmath634 : @xmath635 for all these pairs we have @xmath632 by direct computation . if @xmath623 , we remove most rows and columns common to @xmath636 and @xmath637 and reduce to @xmath638 or @xmath639 , which satisfy @xmath632 by direct computation .
we now assume that we are not in one of the exceptional cases outlined above and so @xmath640 and @xmath641 .
remove all columns common to @xmath636 and @xmath637 to obtain partitions @xmath642 and @xmath643 . in the case
@xmath609 , we have that @xmath644 is a 2-line partition and @xmath645 such that @xmath646
. therefore @xmath632 by proposition [ prop : res2part ] . in the case
@xmath647 , @xmath648 with @xmath640 and @xmath649 and @xmath67 and @xmath578 are @xmath132-removable ; the result follows as @xmath650 .
we now assume that @xmath651 with @xmath652 ; by [ induction ] we can assume that @xmath578 is a fat hook .
we first deal with the exceptional cases in which @xmath624 or @xmath653 .
if @xmath654 then we remove all but one row or column common to both @xmath58 and @xmath135 to obtain pairs of partitions @xmath321 .
for all other pairs of partitions of 4 or 5 , we remove all rows and columns common to both @xmath58 and @xmath135 to obtain @xmath321 .
the partitions @xmath321 obtained in this fashion are all of degree less than or equal to 28 , and so can be checked directly ( one can reduce this degree even further using the semigroup property , but we do not wish to go into these arguments here ) .
we now assume that @xmath651 and @xmath655 .
we remove all rows and columns common to both @xmath39 and @xmath135 to obtain @xmath644 and @xmath643 . if @xmath656 then @xmath657 and so both @xmath658 and @xmath659 are @xmath132-removable ; therefore @xmath660 . if @xmath661 it remains to check each of the possible seven such cases .
if @xmath662 then we may remove an appropriate hook of length @xmath529 from @xmath644 ( namely @xmath663 , respectively ) and the final row of length @xmath529 from @xmath643 to obtain a pair of partitions @xmath664 which differ only by adding and removing a single node ; so the result follows from [ lingamrec ] .
if @xmath665 then @xmath666 and @xmath658 and @xmath659 are both @xmath132-removable ; the result follows as @xmath667 by [ squares ] .
if @xmath668 or @xmath669 then @xmath658 and @xmath659 are both linear and the result follows from [ lingamrec ] . if @xmath670 then @xmath671 and so we are done by [ dvirmaximal ] .
[ animportantremark ] in the proof of [ lingamrec2 ] , we used our assumptions on @xmath58 and @xmath135 to reduce our list of exceptional cases to the pairs @xmath672 and @xmath673 , whereas one naively could have thought we had to check @xmath674 in future proofs , we shall use this technique ( as detailed in the proof above ) without going into further detail .
[ givemealabel ] if @xmath675 or @xmath577 for @xmath640 , then @xmath312 . note that by lemmas [ lingamrec ] and [ lingamrec2 ] , we may assume that @xmath67 is non - linear .
if @xmath676 and @xmath677 then we remove all but possibly one common row or column from @xmath39 and @xmath135 to obtain a pair of partitions @xmath634 equal to one of the seeds @xmath678 or @xmath679 .
we may now assume that @xmath680 is non - linear and @xmath681 .
remove all rows and columns common to both @xmath39 and @xmath135 to obtain pairs of partitions @xmath366 equal to @xmath682 for @xmath665 and @xmath683 , respectively ; here @xmath684 is the rectangular complement of @xmath586 .
let @xmath685 ( and @xmath686 ) for @xmath640
. then @xmath687 and @xmath688 .
the result follows as @xmath357 by [ section2parter ] . if @xmath689 and @xmath665 then @xmath690 and @xmath691 are @xmath132-removable and @xmath692 .
the result follows as @xmath693 by [ squares ] .
if @xmath683 , and @xmath67 is a rectangle such that @xmath689 , then the partitions @xmath680 and @xmath694 are @xmath132-removable and @xmath695 by [ rectan ] . by [ lingamrec2 ] and
the above , we can now assume that @xmath576 is a hook not equal to @xmath615 or @xmath616 . if @xmath696 , then @xmath697 is a proper fat hook and @xmath698 $ ] is not the natural character ; therefore @xmath357 by [ induction ] .
finally , if @xmath699 then @xmath700 we have that @xmath701 by [ lingamrec2 ] .
the result follows by [ prop : monotonicity ] .
[ rectnagledeltalinear ] if @xmath702 then @xmath312 .
we first consider the case where @xmath67 is a rectangle .
if @xmath676 , then we remove almost all rows and columns common to @xmath39 and @xmath135 to obtain @xmath634 equal to one of the seeds @xmath703
@xmath704 @xmath705 .
if @xmath706 for @xmath170 then we remove almost all common rows and columns of @xmath39 and @xmath135 to obtain @xmath707 equal to either of @xmath708 in the former case , the result follows by [ sectionhook ! ] as @xmath319 is a hook . in the latter case the result follows from [ section2parter ] as
@xmath643 is a 2-line partition .
now assume @xmath709 for @xmath170 .
we remove almost all common rows and columns to obtain pairs of partitions @xmath710 equal to either of @xmath711 in the former case , @xmath644 is a 2-line partition and the result follows . in the latter case , remove the final row of @xmath643 and the partition @xmath712 from the first row of @xmath644 to obtain partitions @xmath713 and @xmath714 .
the result again follows from [ section2parter ] .
we now consider the case that @xmath715 is a fat rectangle for @xmath716 .
we may proceed as above by removing all but one common row or column to obtain pairs of partitions @xmath710 equal to either of @xmath717 respectively . in the former ( respectively latter )
case the result follows from [ section2parter ] ( respectively [ sectionhook ! ] ) .
we now assume that @xmath718 is a non - rectangular fat hook , in other words @xmath719 and @xmath720 .
we first consider the case where @xmath721 or @xmath722 . by assumption
, @xmath723 has at least two removable nodes @xmath724 and @xmath725 such that @xmath726 and @xmath578 are disconnected for @xmath727 .
we may assume that @xmath728 is not a rectangular partition .
we want to apply [ lem : prepalmext ] and recall the definition of the virtual character @xmath26 given there in [ chi ] ; note that here @xmath729 .
@xmath730 \cdot [ \mu / \beta_a ] - \sum_{b \in { \operatorname{add}}(\alpha ) } \alpha^b \:.\ ] ] for the two terms on the right - hand side , we note that the subtracted term is multiplicity - free . by assumption @xmath731 = [ k+1]+[k,1]$ ] for @xmath727 .
also note that @xmath732=[{\alpha}^{a_1}]$ ] .
therefore , we have that @xmath733 \rangle & { \geqslant}\sum_{i=1,2 } \langle [ \mu / \beta_{a_i } ] \cdot [ \lambda/ \beta_{a_i } ] , [ { \alpha}^{a_1 } ] \rangle - 1 \\ & { \geqslant}\langle [ \mu / \beta_{a_1 } ] \cdot([k+1 ] + [ k,1 ] ) , [ { \alpha}^{a_1 } ] \rangle + \langle [ \mu / \beta_{a_2 } ] \cdot ( [ k+1 ] + [ k,1 ] ) , [ { \alpha}^{a_1 } ] \rangle -1 \\ & { \geqslant}2 + 1 - 1 \end{aligned}\ ] ] and the result follows by [ lem : prepalmext ] .
we now consider the case in which @xmath734 and @xmath735 and so @xmath736 .
if @xmath737 , then the result follows as @xmath58 is a 2-line partition .
if @xmath738 , we remove the final @xmath739 rows from @xmath135 and @xmath740 from the first row of @xmath39 to obtain @xmath741 ( and so has at least three lines ) and @xmath644 a partition which is neither a hook nor a 2-line partition . in this case , @xmath742 is a rectangle , therefore the result follows from the above and [ prop : monotonicity ] .
finally , we consider the case in which @xmath743 is not a fat hook , i.e. , @xmath744 .
then we apply the following iterative procedure to reduce to the situation dealt with before . *
if @xmath745 , and @xmath744 , then we remove all columns common to both @xmath58 and @xmath135 to obtain a pair @xmath321 such that @xmath746 and therefore @xmath747 , @xmath748 .
* if @xmath749 , and @xmath750 , then we remove the final @xmath751 rows from @xmath135 and the corresponding number of nodes from @xmath752 to obtain a pair @xmath321 such that @xmath753 and @xmath754 and @xmath755 , @xmath756 . *
having completed @xmath757 or @xmath435 above , relabel the partitions @xmath758 and apply @xmath757 or @xmath435 again , if possible .
the above procedure eventually terminates by producing a pair of partitions @xmath759 such that @xmath760 , @xmath761 , @xmath762 ; therefore the result follows by the semigroup property and the case for fat hooks , covered above .
[ lem : delta - nat ] if @xmath763 , then @xmath312 . by [ induction ] , we may assume that @xmath743 is a fat hook .
if @xmath706 and @xmath764 , then we remove all but one row or column of @xmath39 and @xmath135 to obtain partitions @xmath644 and @xmath643 such that @xmath765 or @xmath766 respectively . if @xmath767 , then remove all rows and columns common to both @xmath39 and @xmath135 to obtain partitions @xmath644 and @xmath768 . in either case , we now remove the final row of @xmath582 to obtain @xmath769 and we let @xmath770 denote the partition such that @xmath771 . the partition @xmath769 is a rectangle and @xmath770 is either a proper fat hook or @xmath772 and such that @xmath773 .
the result follows from [ rectnagledeltalinear ] .
[ nonatural ] for the remainder of this section , we shall assume that @xmath774 $ ] is not equal to a linear character or the natural character or its conjugate .
similarly if @xmath578 has one connected component , then we shall assume that @xmath775 $ ] is not equal to a linear character or the natural character or its conjugate .
[ 2erssss ] if @xmath576 and @xmath578 are both 2-line partitions , then @xmath312 .
we first suppose that @xmath776 .
there are three cases to consider : @xmath195 @xmath706 ; @xmath196 @xmath777 and @xmath778 ; @xmath213 @xmath777 and @xmath779 .
case @xmath195 . remove all common rows and all but one common column of @xmath58 and @xmath135 to obtain @xmath780 . for @xmath625 and @xmath633
it is easily checked that the corresponding pairs are seeds .
when @xmath487 , we note that at least one of @xmath781 is @xmath132-removable from @xmath578 ( and hence is also @xmath132-removable from the first two rows of @xmath319 ) . in this case , we remove the final two rows of @xmath582 and the relevant partition from @xmath319 to obtain @xmath782 such that @xmath783 is a non - linear rectangle and @xmath784 is a proper skew partition not of one of the forms described in cases @xmath435 and @xmath411 in [ thm : classification - skew ] .
the result then follows by [ induction ] . in case
@xmath196 ( respectively @xmath213 ) we remove all rows and columns common to @xmath58 and @xmath135 to obtain @xmath582 a 2-column rectangle and @xmath319 a proper fat hook ( respectively @xmath319 such that @xmath785 ) .
the result then follows from [ section2parter ] . for the remainder of the proof
we assume that at least one of @xmath786 and @xmath787 is greater than 2 . in the generic case
, we remove all common rows and columns from @xmath58 and @xmath135 to obtain partitions @xmath319 and @xmath582 and proceed case - by - case .
we will deal with the exceptional cases when they appear in that discussion .
if @xmath788 and @xmath789 ( respectively @xmath706 and @xmath790 ) then @xmath319 ( respectively @xmath582 ) is a 2-line partition and the result follows from [ section2parter ] as long as we are not in the case @xmath791 . in the exceptional cases we remove all but one common row or column from @xmath58 and @xmath135 to obtain @xmath321 . for @xmath788
, we have in the exceptional case @xmath792 or @xmath793 , respectively .
hence we can reduce to @xmath794 or @xmath795 , respectively , and @xmath796 by [ section2parter ] . in the exceptional case for @xmath706
we quickly reduce to a pair involving a 2-column partition where we can again appeal to [ section2parter ] . if @xmath788 and @xmath790 , then @xmath578 and @xmath67 are @xmath132-removable and @xmath797 and the result follows from [ squares ] .
we may now assume @xmath798 up to conjugation .
if @xmath685 , then @xmath582 is a 2-line partition and @xmath319 is a proper fat hook or @xmath785 .
now assume @xmath799 and @xmath789 .
if @xmath800 , remove the two lower rows from @xmath582 to obtain @xmath801 , and note that @xmath802 where @xmath803 is a partition with @xmath804 ; hence the result follows by [ section2parter ] .
if @xmath805 for @xmath806 , we have @xmath807 for some partition @xmath803 , and @xmath808 , so with @xmath809 the claim follows . it remains to check the cases in which @xmath799 and @xmath790 ; namely @xmath195 @xmath810 and @xmath811 ; @xmath196 @xmath812 and @xmath811 ; @xmath213 @xmath813 and @xmath814 ; @xmath520 @xmath813 and @xmath811 ; @xmath815 @xmath816 and @xmath817 . in case @xmath195 , for @xmath655 ( one can check the seeds for @xmath818 directly ) we have that @xmath819 and @xmath820 are @xmath132-removable and the rectangle @xmath821 contains @xmath346 , so the result follows from [ squares ] .
case @xmath196 is similar . in cases @xmath213 to @xmath815 , we have that @xmath819 and @xmath820 are linear partitions and the result follows from [ lingamrec ] .
if @xmath67 or @xmath578 is a fat rectangle and @xmath578 has one connected component , then @xmath312 . by [ induction ] and [ nonatural ]
we may assume that one of @xmath578 and @xmath67 is a fat rectangle and the other is @xmath822 , or @xmath823 or @xmath496 or @xmath497 up to conjugation .
we first suppose that @xmath67 is a fat rectangle .
remove all rows and columns common to both partitions @xmath145 to obtain @xmath644 and @xmath643 .
if @xmath824 , then the result holds as @xmath67 and @xmath578 are @xmath132-removable and @xmath825 . in the remaining cases , @xmath826 , @xmath827 or @xmath497 ,
the partition @xmath319 is a 2-line partition and @xmath582 is a fat rectangle ; the result follows by [ section2parter ] .
we now suppose @xmath578 is a fat rectangle .
remove all rows and columns common to both partitions @xmath145 to obtain @xmath828 . for @xmath829 or @xmath497 ,
this follows via @xmath132-removability from @xmath830 , the conjugate case is immediate from [ section2parter ] .
if @xmath831 or @xmath832 then @xmath821 is a fat rectangle and @xmath833 and @xmath834 are @xmath132-removable ; the result follows by [ squares ] .
if @xmath835 , then @xmath582 is a 2-line partition and @xmath319 is a proper fat hook ; the result follows by [ section2parter ] .
[ hookhook ] if one of @xmath576 or @xmath578 is a hook and the other is equal to @xmath836 or @xmath837 , then @xmath312 .
first note that our assumptions in [ nonatural ] imply that @xmath838 .
if @xmath578 is a hook , remove all columns common to @xmath58 and @xmath135 to obtain @xmath319 and @xmath582 .
if @xmath706 then the result follows from the result for 2-line partitions .
if @xmath788 , then @xmath839 and @xmath840 are @xmath132-removable and @xmath841 is a fat rectangle ; the result follows by [ rectan ] .
now assume that @xmath66 is a hook and @xmath842 or @xmath836 .
remove all rows and columns common to both @xmath145 to obtain partitions @xmath843 and @xmath844 or @xmath845 respectively ; in these cases , @xmath846 and @xmath847 , respectively . in the case
@xmath848 , i.e. , @xmath849 is a square , we must have @xmath811 .
let @xmath850 , where we have @xmath851 .
since @xmath852 and @xmath853 , we obtain @xmath854 .
hence the final @xmath535 rows of @xmath849 form a partition of size @xmath855 ; removing this gives @xmath856 . on the other hand , we can remove a partition of the corresponding size from @xmath319 , as @xmath857 , with @xmath858 . thus @xmath770 and @xmath769 are @xmath132-removable ; since @xmath859 and @xmath860 , we are done in this case .
hence we may now assume that @xmath861 .
we conjugate the partition @xmath643 and consider the possible intersection diagrams @xmath862 and @xmath863 . by our assumptions we have @xmath864 , @xmath865 ,
thus if @xmath867 is disconnected , both components are of size strictly greater than 1 .
when @xmath811 , @xmath867 could be disconnected only when both @xmath868 and @xmath869 , which is impossible .
when @xmath778 , @xmath867 is disconnected if and only if @xmath870 .
then @xmath871 is a rectangle of width @xmath872 and height @xmath873 . if @xmath874 , then @xmath875 , hence @xmath876 , contradiction .
hence @xmath871 is a non - linear rectangle .
but then the character pair @xmath877,[d_2])$ ] is not on the list in [ thm : classification - skew ] ; hence by induction @xmath878 , and thus @xmath312 .
so we now assume that @xmath879 are both connected ; in fact , then @xmath867 must be a rectangle and @xmath871 is a fat hook . if @xmath871 is a proper fat hook or @xmath867 a fat rectangle , or if one is a rectangle and the other is not a hook , we are done by the previous results of this section .
it remains to consider the case where @xmath867 is a 2-line rectangle of size @xmath880 and @xmath871 is a hook of the form @xmath881 for @xmath882 .
when @xmath778 , this implies that @xmath874 and @xmath883 , hence @xmath624 , @xmath884 .
when @xmath811 , this implies that @xmath885 and @xmath886 , so again @xmath624 , @xmath884 . in both cases ,
we can remove a column of length 8 from @xmath582 and remove @xmath887 from @xmath319 , and the result then follows from [ lingamrec2 ] . if @xmath578 has two connected components , then @xmath312 . by [ induction ] , it suffices to consider ( up to conjugation of @xmath58 and @xmath135 ) the cases @xmath195 @xmath888 or @xmath889 and @xmath890 and @xmath891 is a rectangle ; @xmath196 @xmath892 and @xmath774=[\delta']\boxtimes [ \delta'']$ ] is one of the products from the list in [ thm : mf - outer ] with @xmath893 of size @xmath894 , respectively , and @xmath895 not a pair as in @xmath195 .
we cover both cases uniformly .
the unique exceptional subcase is @xmath896 and @xmath897 , @xmath898 ( up to conjugation of @xmath58 and @xmath135 ) in which case we remove all rows and columns common to both partitions with the exception of one row ( which exists by our assumption that @xmath135 is not a 2-line partition ) to obtain @xmath321 of the form @xmath899 now suppose that @xmath896 and @xmath900 and @xmath901 are not of the above form .
remove all rows and columns common to @xmath58 and @xmath135 to obtain @xmath319 and @xmath582 .
now , if @xmath902 and @xmath903 then @xmath904 is disconnected , with two components @xmath905 and @xmath757 , and @xmath906 ; for @xmath907 the result follows as this product is not on the list in [ thm : classification - skew ] , and for @xmath908 , the pair @xmath909 is a seed .
if @xmath910 ( in either the exceptional or generic cases ) then @xmath582 is a rectangle and @xmath911 and therefore @xmath912 and @xmath913 each have precisely one component .
the result follows by the earlier results in this section .
we now suppose that @xmath902 and @xmath914 or @xmath915 for some @xmath916 and that @xmath715 for @xmath917 .
remove all rows and columns common to @xmath58 and @xmath135 to obtain @xmath319 and @xmath582 .
if @xmath918 , then @xmath919 is a hook partition and @xmath582 is a fat rectangle and the result holds by [ sectionhook ! ] .
if @xmath888 , then we remove the first row @xmath920 of @xmath319 and the final @xmath739 columns of @xmath582 to obtain the pair @xmath921 and the result follows by [ squares ] .
in this section , we shall consider tensor products in which one of the labelling partitions is a _ proper _ fat hook ( in other words , a fat hook which is not a 2-line , hook , or rectangular partition ) . we assume throughout this section that @xmath922 is a proper fat hook partition .
let @xmath73 .
the product @xmath205 \cdot [ { \lambda}]$ ] is multiplicity - free if and only if @xmath923 or @xmath8 up to conjugation .
one half of the proposition follows from [ sec : mf - products ] . in this section , we prove the other half of this proposition via a series of lemmas . for the remainder of this section , we assume that @xmath924 or @xmath8 up to conjugation and we want to deduce that @xmath5 \cdot [ \mu]$ ] contains multiplicities .
we may assume that @xmath39 is neither a rectangle , a hook , or 2-line partition , and that @xmath557 , as we have already dealt with these cases in [ sec : warm - up , sec : rectangle ] .
the possible intersection diagrams for @xmath39 and @xmath135 , up to conjugation , are given in [ singlefatty , singlefatty2 ] .
we will also use the notation indicated there , in other words we let @xmath559 , @xmath560 and @xmath925 .
informally , we refer to the overlapping rectangles of shape @xmath515 and of shape @xmath926 as the arm and the leg of @xmath135 , respectively .
@xmath927 \draw[very thick ] ( -1,1)--(4,1)--(4,-2)--(2,-2)--(2,-5)--(-1,-5)--(-1,1 ) ; \draw ( 4,1)--(8,1)--(8,0.5)--(7,0.5)--(7,0)--(6.5,0)--(6.5,-0.5)--(4,-0.5)--(4,1 ) ; \draw ( 4,-1)--(3,-1)--(3,-1.5)--(1.5,-1.5)--(1.5,-2.5)--(1,-2.5)--(1,-3)--(0.5,-3)--(0.5,-5 ) ; \draw(1.25,-3.5 ) node { $ \gamma$ } ; \draw(5.5,0.25 ) node { $ \delta$ } ; \end{tikzpicture}\end{minipage } \begin{minipage}{32mm}\begin{tikzpicture}[scale=0.4 ] \draw[very thick ] ( -2,1)--(4,1)--(4,-2)--(2,-2)--(2,-5)--(-2,-5)--(-2,1 ) ; \draw ( 4,1)--(5,1)--(5,0.5)--(4.5,0.5)--(4.5,-2.5)--(4,-2.5)--(4,-3)--(2,-3 ) ; \draw ( 2,-3.5)--(.5,-3.5)--(.5,-4)--(-1,-4)--(-1,-5 ) ; \draw(1,-4.5 ) node { $ \gamma$ } ; \draw(3,-2.5 ) node { $ \delta$ } ; \end{tikzpicture}\end{minipage } \begin{minipage}{37mm}\begin{tikzpicture}[scale=0.4 ] \draw[very thick ] ( -1,1)--(4,1)--(4,-2)--(2,-2)--(2,-5)--(-1,-5)--(-1,1 ) ; \draw ( 4,1)--(7,1)--(7,0.5)--(6,0.5)--(6,0)--(5.5,0)--(5.5,-0.5)--(4,-0.5)--(4,1 ) ; \draw ( 2,-3.5)--(1.5,-3.5)--(1.5,-4)--(0,-4)--(0,-5 ) ; \draw(1,-4.5 ) node { $ \gamma$ } ; \draw(5,0.25 ) node { $ \delta$ } ; \end{tikzpicture}\end{minipage } \begin{minipage}{24mm}\begin{tikzpicture } [ scale=0.4,x={(0,-1cm)},y={(-1cm,0 ) } ] \draw[very thick ] ( 0,1)--(4,1)--(4,-2)--(1,-2)--(1,-3)--(-0,-3)--(0,1 ) ; \draw ( 4,1)--(6.5,1)--(6.5,0.5)--(6,0.5)--(6,0)--(5.5,0)--(5.5,-0.5)--(4,-0.5)--(4,1 ) ; \draw ( 4,-1)--(2,-1)--(2,-1.5)--(1.5,-1.5)--(1.5,-2 ) ; \draw(3,-1.5 ) node { $ \gamma$ } ; \draw(5,0.25 ) node { $ \delta$ } ; \end{tikzpicture}\end{minipage } \begin{minipage}{35mm}\begin{tikzpicture}[scale=0.4,x={(0,-1cm)},y={(-1cm,0 ) } ] \draw[very thick ] ( -0.5,1)--(4,1)--(4,-2)--(0.5,-2)--(0.5,-5)--(-0.5,-5)--(-0.5,1 ) ; \draw ( 4,0.5)--(3.5,0.5)--(3.5,-0.5)--(3,-0.5)--(3,-0.5)--(3,-1)--(3,-1.5)--(2.5,-1.5)--(2.5,-2 ) ; \draw(3.5,-1.25 ) node { $ \gamma$ } ; \draw ( 2,-2)--(2,-2.5)--(1.5,-2.5)--(1.5,-4)--(1,-4)--(1,-4.5)--(0.5,-4.5 ) ; \draw(1,-3.25 ) node { $ \delta$ } ; \end{tikzpicture}\end{minipage}\ ] ] @xmath928 \draw[very thick ] ( -1,1)--(4,1)--(4,-2)--(1,-2)--(1,-3.5)--(-1,-3.5)--(-1,1 ) ; \draw ( 4,1)--(6,1)--(6,0.5)--(5,0.5)--(5,0)--(4.5,0)--(4.5,-0.5)--(4,-0.5)--(4,1 ) ; \draw ( 4,-1)--(2,-1)--(2,-1.5)--(1.5,-1.5)--(1.5,-2 ) ; \draw ( 0.5,-3.5)--(0.5,-4)--(0,-4)--(0,-4.5)--(-0.5,-4.5)--(-0.5,-5)--(-1,-5)--(-1,-3 ) ; \draw(3,-1.5 ) node { $ \gamma$ } ; \draw(4.6,0.5 ) node { $ \delta''$ } ; \draw(-0.5,-4 ) node { $ \delta'$ } ; \end{tikzpicture}\end{minipage } \begin{minipage}{32mm}\begin{tikzpicture } [ scale=0.4,x={(0,-1cm)},y={(-1cm,0 ) } ] \draw[very thick ] ( -1,1)--(4,1)--(4,-2)--(0,-2)--(0,-4.5)--(-1,-4.5)--(-1,1 ) ; \draw ( 4,1)--(6,1)--(6,0.5)--(5,0.5)--(5,0)--(4.5,0)--(4.5,-0.5)--(4,-0.5)--(4,1 ) ; \draw ( 4,-1)--(2,-1)--(2,-1.5)--(1.5,-1.5)--(1.5,-2 ) ; \draw ( 1,-2)--(1,-3.5)--(0.5,-3.5)--(0.5,-4)--(0,-4 ) ; \draw(3,-1.5 ) node { $ \gamma$ } ; \draw(4.6,0.5 ) node { $ \delta''$ } ; \draw(0.5,-3 ) node { $ \delta'$ } ; \end{tikzpicture}\end{minipage } \begin{minipage}{32mm}\begin{tikzpicture}[scale=0.4 ] \draw[very thick ] ( -1,1)--(4,1)--(4,-2)--(2,-2)--(2,-5)--(-1,-5)--(-1,1 ) ; \draw ( 4,1)--(5.5,1)--(5.5,-0)--(4.5,-0)--(4.5,-0.5)--(4.5,-0.5)--(4.5,-1.5)--(4,-1.5)--(4,-2)--(3.5,-2)--(3.5,-2.5)--(3,-2.5)--(3,-3)--(3,-3)--(2,-3 ) ; \draw ( 2,-3.5)--(.5,-3.5)--(.5,-4)--(-0.5,-4)--(-0.5,-4)--(-0.5,-5 ) ; \draw(1,-4.5 ) node { $ \gamma$ } ; \draw(2.65,-2.5 ) node { $ \delta''$ } ; \draw(5,0.5 ) node { $ \delta'$ } ; \draw[white ] ( 5,-5.25 ) node { $ \delta$ } ; \end{tikzpicture}\end{minipage } \begin{minipage}{32mm}\begin{tikzpicture}[scale=0.4 ] \draw[very thick ] ( -2,0)--(4,0)--(4,-2)--(2,-2)--(2,-5)--(-2,-5)--(-2,0 ) ; \draw ( 4,0)--(5,0)--(5,-0.5)--(4.5,-0.5)--(4.5,-2.5)--(4,-2.5)--(4,-3)--(2,-3 ) ; \draw ( 2,-3.5)--(.5,-3.5)--(.5,-4)--(0,-4)--(0,-4)--(0,-5 ) ; \draw ( -0.5,-5)--(-0.5,-5.5)--(-1,-5.5)--(-1,-6)--(-2,-6)--(-2,-5 ) ; \draw(1,-4.5 ) node { $ \gamma$ } ; \draw(3,-2.5 ) node { $ \delta'$ } ; \draw(-1.5,-5.5 ) node { $ \delta''$ } ; \end{tikzpicture}\end{minipage}\ ] ] @xmath929 \draw[very thick ] ( -1,1)--(4,1)--(4,-2)--(2,-2)--(2,-5)--(-1,-5)--(-1,1 ) ; \draw ( 4,1)--(7,1)--(7,0.5)--(6,0.5)--(6,0)--(5.5,0)--(5.5,-0.5)--(4,-0.5)--(4,1 ) ; \draw ( 2,-3.5)--(1.5,-3.5)--(1.5,-4)--(0,-4)--(0,-5 ) ; \draw ( 4,-1)--(3,-1)--(3,-1.5)--(2.5,-1.5)--(2.5,-2 ) ; \draw(1,-4.5 ) node { $ \gamma'$ } ; \draw(3.5,-1.5 ) node { $ \gamma''$ } ; \draw(5,0.25 ) node { $ \delta$ } ; \end{tikzpicture}\end{minipage } \begin{minipage}{34mm}\begin{tikzpicture}[scale=0.4 ] \draw[very thick ] ( -1,1)--(7,1)--(7,-1.5)--(2,-1.5)--(2,-5)--(-1,-5)--(-1,1 ) ; \draw ( 3.5,-3 + 1.5)--(3.5,-3 + 0.5)--(3,-3 + 0.5)--(3,-3 + 0)--(3,-3 + 0)--(3,-2.5 + -0.5)--(-2 + 4,-2.5 + -0.5)--(-2 + 4,-3 + 1 ) ; \draw ( 2,-3.5)--(1.5,-3.5)--(1.5,-4)--(0,-4)--(0,-5 ) ; \draw ( 7,0)--(6,0)--(6,-0.5)--(4,-0.5)--(4,-1.5 ) ; \draw(1,-4.5 ) node { $ \gamma'$ } ; \draw(2.7,-2.17 ) node { $ \delta$ } ; \draw(6.25,-0.75 ) node { $ \gamma''$ } ; \end{tikzpicture}\end{minipage}\ ] ] [ fathooklemma2rec ] if @xmath67 and @xmath578 are both rectangles , then @xmath312 .
we invite the reader to check the cases where the size of the partitions @xmath67 and @xmath578 is at most 2 by hand .
these can easily be reduced to small cases ( however , listing them is a somewhat tedious exercise ) .
one can easily show that these products contain multiplicities using simplifications of the arguments used here ( or these can be checked by computer as the degrees of the partitions are small ) .
we shall assume throughout that @xmath67 , @xmath578 are of size strictly greater than 2 .
assuming @xmath67 and @xmath578 are both rectangles , cases @xmath930 and @xmath931 are empty .
we first consider case @xmath932 . if @xmath933 and and @xmath934 ( and by assumption @xmath935 and either @xmath936 or @xmath937 is strictly greater that 1 ) we remove all but one column in the arm and all but one row common to both partitions and hence arrive at @xmath938 equal to one of the following subcases @xmath939
if @xmath933 and @xmath940 ( and by assumption @xmath935 and either @xmath936 or @xmath937 is strictly greater than 1 ) we remove most rows and columns in order to arrive at @xmath941 in all four of the above subcases , we have that @xmath942 by [ section2parter , sectionhook ! ] .
now assume that @xmath943 and @xmath944 , with @xmath910 . by [ animportantremark ] and
our assumption that @xmath945 , we can remove all rows and all but two of the columns common to both partitions until we obtain @xmath321 equal to one of the following subcases @xmath946 the latter case follows by [ section2parter ] . in the former case ,
remove the final two rows of @xmath319 and the final column of @xmath582 to obtain @xmath947 ; the result follows by [ section2parter ] . by [ animportantremark ] , if @xmath948 we can reduce using the semigroup property to the seed @xmath949 .
we now consider the generic case ( not of the above form ) for @xmath932 .
remove all rows and all columns common to both @xmath58 and @xmath135 with the exception of one column from the arm .
we hence obtain @xmath950 . if @xmath951 and @xmath952 then the result follows from [ section2parter ] ( the @xmath953 case was covered above ) .
if @xmath954 and @xmath789 then the result follows from [ section2parter ] .
if @xmath954 and @xmath790 then @xmath67 and @xmath578 are @xmath132-removable and @xmath955 and the result follows by [ squares ] .
we now consider case @xmath956 ; there are two subcases . if @xmath957 , then we remove all common columns from the arms of @xmath135 and @xmath39 until we obtain the partitions @xmath958 and @xmath642 a proper fat hook .
if @xmath665 , then we remove common columns and rows until we obtain @xmath959 and @xmath960 .
the result follows by [ section2parter , sectionhook ! ] .
we now consider case @xmath961 . by [ animportantremark ] , if @xmath962 we can remove successive rows and columns from @xmath135 and @xmath39 until we obtain @xmath366 equal to one of the following pairs : @xmath963 the first ( respectively second ) case follows by [ sec : rectangle ] ( respectively [ squares ] ) . by [ animportantremark ] , if @xmath964 and @xmath665 we can remove all but one row in the arm and all but one common column to obtain @xmath965 equal to one of the pairs : @xmath966 the latter case follows by [ section2parter ] .
the former can be further reduced to the seed @xmath967 by [ animportantremark ] , if @xmath968 and @xmath969 we can successively remove common rows and columns until we obtain @xmath965 equal to one of the following pairs : @xmath970 and the result follows by [ sec : rectangle ] and [ section2parter ] respectively .
we now consider the generic case for @xmath961 .
remove all rows and columns common to both @xmath135 and @xmath39 with the exception of one row in the arm to obtain @xmath971 a proper fat hook and @xmath972 a non - linear rectangle .
the result follows by [ sec : rectangle ] .
[ 7.2 ] if either @xmath67 or @xmath578 is linear and the other is connected , then @xmath312 .
we may assume that one diagram is linear and the other is not a rectangle , as the case of two rectangles has already been addressed in [ fathooklemma2rec ] .
we first consider cases @xmath973 with @xmath67 a linear partition .
suppose we are in case @xmath932 with @xmath934 . if @xmath789 , remove all rows and columns common to @xmath58 and @xmath135 with the exception of one column in the arm to obtain @xmath582 a proper fat hook and @xmath974 a 2-line partition .
hence the result follows by [ section2parter ] .
if @xmath790 remove all rows and columns common to both @xmath135 and @xmath39 to obtain @xmath975 such that @xmath976 a rectangle and @xmath977 .
the result follows by [ sec : rectangle ] .
now consider case @xmath932 with @xmath940 .
remove all rows and columns shared by @xmath135 and @xmath39 with the exception of one column in the arm to obtain @xmath582 and @xmath319 such that @xmath978 is a 2-line partition and @xmath979 is a non - rectangular partition .
the result follows by [ section2parter ] .
for case @xmath956 with @xmath67 linear , we remove all rows and columns common to both @xmath58 and @xmath135 with the exception of one row from the arm to obtain @xmath582 a non - linear rectangle and @xmath319 such that @xmath980 .
the result follows by [ sec : rectangle ] . in case @xmath961 and @xmath934 with @xmath487 , remove all rows and columns common to both @xmath135 and @xmath39 to obtain @xmath981 a rectangle and @xmath979 a non - rectangular partition ; for @xmath633 we reduce to the seed @xmath982 . in case @xmath961 and @xmath940 , remove all rows and columns common to both @xmath58 and @xmath135 with the exception of one row in the arm to obtain @xmath983 .
we have that @xmath984 is a hook partition and @xmath985 ; the result follows by [ sectionhook ! ] .
we now consider cases @xmath973 for @xmath578 a linear partition .
recall that @xmath986 .
assume we are in case @xmath987 with @xmath988 .
if @xmath933 for case @xmath987 and @xmath989 , remove all rows and columns common to @xmath58 and @xmath135 to obtain @xmath582 a fat rectangle and @xmath319 such that @xmath990 and @xmath319 is of depth at least 3 . therefore the result follows by [ induction ] for cases @xmath987 with @xmath933 , @xmath989 , and @xmath988 .
continuing with case @xmath961 with @xmath933 , we now assume that either @xmath685 or @xmath991 . in either case , remove all rows and columns common to both @xmath58 and @xmath135 with the exception of one row in the arm to obtain a pair of proper fat hooks of the form @xmath992 we have that @xmath993 by our assumption that @xmath685 or @xmath991 . by the semigroup property
, we can reduce to @xmath994 and the result follows by [ section2parter ] . continuing with case @xmath932 with @xmath933
, we now assume that either @xmath685 or @xmath991 .
remove all rows and columns common to @xmath58 and @xmath135 with the exception of either one arbitrary row , or one column in the leg .
we hence obtain @xmath582 a rectangle and @xmath319 with at least three removable nodes ; the result follows by [ sec : rectangle ] .
we now consider cases @xmath987 with @xmath665 .
remove all rows and columns common to both partition to obtain @xmath582 a non - linear rectangle and @xmath995 ; the result follows by [ sec : rectangle ] . for case @xmath956 with @xmath578 linear , remove all columns and all but one row common to both @xmath58 and @xmath135 to obtain @xmath582 a fat rectangle and @xmath319 a partition with at least 3 removable nodes .
the result follows by [ sec : rectangle ] .
we now consider case @xmath931 ; here , only @xmath67 can be linear . if @xmath934 remove all common rows and columns to obtain @xmath582 and @xmath319 .
if @xmath789 then the result follows by [ section2parter ] .
suppose that @xmath952 .
the shortest row of @xmath582 is longer than the longest column in @xmath319 and so @xmath996 is a rectangle . by assumption ,
@xmath324 and so @xmath997 and the result follows by [ squares ] .
we now consider the case @xmath931 with @xmath940 .
the exceptional cases are @xmath195 @xmath998 and @xmath196 @xmath789 . in either case , remove all rows and columns with the exception of one column in the leg ( which exists by assumption that @xmath135 is neither a hook , nor a 2-line partition ) to obtain @xmath321 .
we have that @xmath999 by assumption and so we can remove the final @xmath1000 columns of @xmath582 and the final @xmath1001 rows of @xmath319 to obtain @xmath1002 and @xmath1003 .
the result follows by [ section2parter ] .
now suppose we are in case @xmath931 with @xmath1004 and we are not in one of the exceptional cases @xmath195 and @xmath196 above . remove all common rows and columns from @xmath135 and @xmath58 to obtain @xmath582 and @xmath319 .
if @xmath582 is a 2-column partition , the result follows .
otherwise , remove all nodes in @xmath319 to the right of the final column of @xmath582 and remove the corresponding number of nodes from the first column of @xmath1005 to obtain a pair @xmath1006 .
we have that @xmath1007 is a proper partition and @xmath1008 is linear .
the result follows from the case @xmath961 for @xmath1009 a proper partition , above .
finally , suppose we are in case @xmath930 ; here only @xmath578 can be linear .
if @xmath67 is a proper partition , remove all common rows ( or all common columns , respectively ) from @xmath58 and @xmath135 to obtain @xmath319 and @xmath582 .
the partitions @xmath1010 and @xmath1011 are now as in case @xmath961 ( respectively @xmath932 ) above and therefore @xmath312 .
it remains to consider the case when @xmath67 is a proper skew partition .
case @xmath195 .
if @xmath1012 and @xmath933 , remove all but one row ( in the arm ) or one column ( in the leg ) to obtain a pair of partitions @xmath321 . in the former case , we remove successive rows from @xmath582 ( and the corresponding number of nodes from the first row of @xmath319 ) until we obtain @xmath582 a fat rectangle and @xmath319 such that @xmath785 .
the result follows by [ sec : rectangle ] . in the latter case ,
remove the final row of @xmath582 and the corresponding number of nodes from the first row of @xmath319 to obtain a pair @xmath1013 . if @xmath769 is a rectangle the result follows . otherwise , @xmath1014 is a proper skew partition and @xmath1015=[k']\boxtimes [ 1]$ ] with @xmath1016 , and the result follows from [ induction ] .
case @xmath196 .
now assume @xmath1017 and @xmath933 and remove all rows and columns common to @xmath58 and @xmath582 to obtain @xmath319 and @xmath582 . if @xmath319 is a hook or 2-line partition the result follows . otherwise , if @xmath1018 for @xmath1019 ( respectively 2 ) remove all rows in @xmath582 which occur below the final row of @xmath319 and remove the corresponding number of nodes from the first row of @xmath319 to obtain @xmath803 and @xmath1020 .
we have that @xmath1021 ( respectively 3 ) and @xmath769 is either a rectangle or a fat hook such that @xmath1022 is a proper partition ( respectively @xmath1023 for @xmath1024 ) . in the former case ,
the result follows either from [ sec : rectangle ] or from noting that @xmath1013 are as in case @xmath930 for @xmath67 a proper partition . in the latter case ,
repeat the above argument for case @xmath195 or case @xmath196 as appropriate .
finally assume @xmath665 in case @xmath930 .
remove all rows and columns common to both @xmath58 and @xmath135 to obtain a pair @xmath321 .
if @xmath685 , then @xmath319 is a proper fat hook and @xmath582 is a 2-line partition and so the result follows by [ section2parter ] . otherwise , by our assumptions @xmath640 and @xmath1025 .
the shortest column of @xmath319 ( which is of length equal to @xmath621 ) is longer than the widest row of @xmath582 ( equal to @xmath786 ) and so @xmath1026 and so the result follows by [ squares ] . [
42879342578837495278354278354 ] if either @xmath775 $ ] or @xmath774 $ ] is equal to @xmath1027\boxtimes [ 1]$ ] up to conjugation , then @xmath312 .
if @xmath775 $ ] or @xmath774 $ ] is of the form @xmath1028 \boxtimes [ k-1]$ ] up to conjugacy , we may assume that the other is a rectangle by [ induction ] .
it is easy to see that case @xmath1029 is never of this form .
we first consider the pairs of partitions @xmath321 which form our exceptional cases , in which it is not possible to remove all rows and columns common to both partitions @xmath145 . in case
@xmath1030 , suppose that @xmath67 is linear and @xmath774 = [ 1 ] \boxtimes [ k-1]$ ] . by [ animportantremark ] , we can remove most rows and columns common to @xmath135 and @xmath39 to obtain @xmath366 equal to one of the following @xmath1031 otherwise , remove all rows and columns common to both @xmath58 and @xmath135 to obtain @xmath321 .
the result follows by [ sectionhook!,sec : rectangle ] .
we now consider case @xmath1032 ( in case @xmath1033 one can use the semigroup property to reduce to the same set of cases , and we therefore do not consider this case explicitly ) .
suppose that @xmath67 is linear and @xmath774 = [ 1 ] \boxtimes [ k-1]$ ] up to conjugation .
the exceptional cases are precisely those in which @xmath1034 ( with notation as in case @xmath1032 of [ singlefatty2 ] ) and @xmath67 is linear .
remove all rows and columns common to both @xmath135 and @xmath39 with the exception of one row in the arm to obtain @xmath366 equal to one of the following up to conjugation @xmath1035 @xmath1036 in each case we can remove a single node from the first row of @xmath582 and a single node from the first column of @xmath319 to obtain a pair @xmath782 . in the first case
@xmath1037 and @xmath803 is a proper fat hook and the result follows by [ section2parter ] . in the third case @xmath769 is a fat rectangle and @xmath1038 is a proper fat hook and the result follows by [ section2parter ] . in the second and fourth cases with @xmath838
( the @xmath625 cases are covered by the first and third cases ) @xmath1039 are both proper fat hooks and @xmath1040 is linear and @xmath1041 $ ] is the standard character and so the result follows by [ 7.2 ] .
the only exceptional case for @xmath1042 is when @xmath1043 and @xmath940 .
then remove all rows and columns common to both @xmath58 and @xmath135 with the exception of one column in the arm or leg ( which must exist as @xmath135 is not a 2-line partition ) to obtain @xmath319 and @xmath582 . in the former case
@xmath319 is a proper fat hook and @xmath1044 is a hook partition ; the result then follow by [ sectionhook ! ] . in the latter case remove a single node from the first column of @xmath319 and the first row of @xmath582 ; the result then follows by [ section2parter ] .
the only exceptional case for @xmath1045 is that in which @xmath1046 and @xmath933 . remove all rows and columns common to both @xmath58 and @xmath135 with the exception of one row or column in the arm or in the leg to obtain @xmath319 and @xmath582 . for a single row in the leg
( respectively column in the arm ) the result then follows by [ sectionhook ! ] ( respectively [ section2parter ] ) . for a single row in the arm ,
remove a single node from the first column of @xmath319 and the first column of @xmath582 ; the result then follows by [ section2parter ] .
for a single column in the leg , remove the final row of @xmath582 and the final two columns of @xmath319 to obtain @xmath1013 . for @xmath838 , both @xmath1047 and @xmath1048 have two connected components and the result follows by [ induction ] ; for @xmath625 , we have the seed @xmath1049 .
now suppose that we are in one of the cases @xmath1050 and @xmath1051 is not one of the exceptional cases ( all of which were dealt with above ) . in cases @xmath1052 we remove all rows and columns common to both @xmath135 and @xmath39 to obtain a pair @xmath321 where @xmath582 or @xmath319 is a proper rectangle , and which is not on our list . in case @xmath1042
we remove all common rows and columns from @xmath145 and obtain either a 2-line partition @xmath319 with @xmath321 not on our list , or a pair which can be reduced in one further step to a pair not on our list where at least one is a proper rectangle . in case @xmath1045 again remove all common rows and columns and obtain either a 2-line partition @xmath319 with @xmath321 not on our list , or a pair @xmath321 where we can remove a shape corresponding to @xmath578 from @xmath319 and the final boxes from the @xmath621 columns of @xmath582 , and @xmath1053 .
so the result follows from [ section2parter ] , [ squares ] and [ sec : rectangle ] . [
7.4 ] if either @xmath67 or @xmath578 is a proper hook partition up to rotation , then @xmath312 . by [ sec : warm - up ] and [ dvirmaximal ] we may assume that up to rotation one of @xmath1054 is a proper hook and the other is a fat hook . by [ 7.2 ] we may assume that neither partition is linear .
assume @xmath325 are as in @xmath930 of [ singlefatty ] .
we begin with the case @xmath1055 ( the case @xmath683 is identical ) .
for @xmath633 we remove all rows and columns common to both partitions with the exception of one column in the leg to obtain the seed @xmath1056 . for @xmath487 , we remove all rows and columns common to both @xmath145 to obtain @xmath366 .
if @xmath1057 , then the result follows from [ section2parter ] . if @xmath1058 , then @xmath67 and @xmath578 are @xmath132-removable and @xmath955
now assume that @xmath1059 , with @xmath487 .
if @xmath621 is even and @xmath1060 , then we remove all rows and columns common to both partitions with the exception of one column in the leg to obtain @xmath582 and @xmath319 .
we then remove @xmath1061 from the top of @xmath319 and @xmath1062 from the bottom of @xmath582 to obtain @xmath1063 and then the result follows by [ 7.2 ] . for @xmath578 not of the above form , remove all rows and columns common to both partitions to obtain @xmath321 such that @xmath582 is a 2-line partition and @xmath366 is not a pair listed in [ thm : classification ] ; the result follows by [ section2parter ] .
if @xmath1064 and @xmath67 is not of the above form , remove all rows and columns common to @xmath58 and @xmath135 to obtain @xmath1065 and @xmath582 a proper fat hook .
the result follows from [ section2parter ] .
we now assume that @xmath1066 up to conjugation . by [ sec
: warm - up ] and [ dvirmaximal ] , being in case @xmath930 implies that @xmath67 is a proper hook and @xmath578 is a non - linear rectangle .
remove all rows and columns common to @xmath135 and @xmath39 to obtain @xmath582 a proper fat hook and @xmath319 a rectangular partition ; the result follows then by [ sec : rectangle ] . before addressing case @xmath931 of [ singlefatty ] ,
we first consider cases @xmath973 .
assume that @xmath578 is a proper hook . in case @xmath932
, there is a single exceptional subcase , where @xmath706 and @xmath764 ; here we remove all rows and columns common to both partitions with the exception of one column in the arm to obtain @xmath1067 ; the result follows by [ section2parter ] . in case @xmath956 , the unique exceptional subcase is @xmath1068 , which we can reduce to the seed @xmath1069 . in case @xmath961 ,
the single exceptional subcase is given by @xmath706 and @xmath764 ; remove all rows and columns common to both partitions with the exception of one row in the arm to obtain @xmath321 . in which case
@xmath1070 is equal to a pair of partitions as in the second exceptional case for @xmath930 , above .
continuing with @xmath973 with @xmath1071 a proper hook , we now argue for the generic case .
remove all rows and columns common to @xmath58 and @xmath135 to reduce to a pair of partitions @xmath321 such that @xmath582 is a rectangle and @xmath321 does not belong to our list . thus the result follows by [ sec : rectangle ] .
suppose that we are in cases @xmath973 and that @xmath67 is a rotated proper hook .
remove all rows and columns common to both partitions to obtain @xmath1072 such that @xmath582 is a rectangle and @xmath319 is a proper fat hook or has three removable nodes ( as @xmath578 is non - linear ) .
the result follows from [ sec : rectangle ] .
finally , we consider case @xmath931 . if @xmath1073 is a proper hook and @xmath67 is a non - linear rectangle , remove all rows and columns common to @xmath145 to obtain @xmath1072 such that @xmath319 is a rectangle and @xmath582 is a proper fat hook ; the result follows then by [ sec : rectangle ] .
we may now assume that one of @xmath576 or @xmath1073 is equal to @xmath616 up to conjugation and the other is a non - rectangular fat hook .
this case is symmetric in swapping @xmath67 and @xmath578 and therefore we can assume that @xmath1074 up to conjugation and @xmath1075 is not a rectangle . remove all rows and columns common to @xmath145 to obtain @xmath321 equal to either @xmath1076 the @xmath633 case is the seed @xmath1077 . for @xmath487 in the latter case , if @xmath576 is of depth at least @xmath1078 , then @xmath1079 and so @xmath1080 and @xmath1081 are @xmath132-removable and @xmath1082 . if the depth of @xmath576 is smaller than @xmath1078 , and @xmath1083 then the sum of the first and final columns in @xmath319 is equal to the sum of the first and final columns in @xmath582 ( equal to @xmath1084 in both cases ) . now
if @xmath576 is not one of @xmath1085 , we remove these columns and obtain @xmath1086 a non - linear rectangle , and @xmath803 a proper fat hook ; the result follows by [ semigroup , sec : rectangle ] .
if @xmath1087 , we reduce to the seed @xmath1088 .
if @xmath1089 , we remove the final two rows from @xmath319 and the final row from @xmath582 , giving a rectangle and a proper fat hook ; the result follows by [ semigroup , sec : rectangle ] .
if @xmath1090 , we remove the final two columns from @xmath319 and the first column from @xmath582 , giving a pair of 2-line partitions not on our list , so the result follows . in the former case with @xmath487 ,
remove the final column of @xmath319 and the final two columns from @xmath582 to obtain @xmath803 and @xmath769 such that @xmath1091 and @xmath1092 . by [ induction ] , if @xmath576 is not equal to @xmath616 up to conjugation , we are done .
if @xmath1093 then @xmath357 by [ squares ] and if @xmath651 , then the result follows by conjugating to the latter case , discussed above . if either @xmath67 or @xmath578 is a 2-line partition , then @xmath312 . by [ induction ] and the previous results in this section , it will suffice to consider @xmath680 and @xmath1094 such that up to conjugation * one is equal to @xmath15 and the other is @xmath1095 ; * the pair is equal to one of the special pairs @xmath18 or @xmath19 ; * one is equal to @xmath1096 and the other is a rectangle ; * the pair is equal to one of @xmath1097 , @xmath1098 , or @xmath1099 .
we consider the proper skew partition case first .
we assume without loss of generality that we are in case @xmath930 ( case @xmath931 is identical and such a pair @xmath67 and @xmath578 can not occur in cases @xmath973 ) .
remove all rows common to @xmath58 and @xmath135 to obtain @xmath321 equal to one of @xmath1100 in the first case , we have that @xmath1101 and @xmath1102 are both linear and so the claim follows from [ 7.2 ] . in the second case ,
we have that @xmath1103 and @xmath1104 and so the result follows for @xmath487 by [ induction ] ; at @xmath625 we have the seed @xmath1105 , to which we also easily reduce in the case @xmath633 .
the third and fourth cases follow from [ section2parter ] . for the remainder of the proof , we assume that @xmath67 and @xmath578 are both proper partitions ( up to rotation ) and proceed case - by - case through @xmath930 to @xmath961 .
we first consider case @xmath930 depicted in [ singlefatty ] . by the above
, @xmath67 is a ( non - rectangular ) 2-line partition .
if @xmath67 is equal to @xmath1106 up to conjugation , then we can remove rows and columns common to @xmath135 and @xmath39 until we are in one of the four cases in [ hellodleo ] or at one of @xmath1107 or @xmath1108 . in the first three cases in [ hellodleo ] ,
the result follows by [ squares , section2parter ] . in the fourth case in [ hellodleo ] , we remove the final row of @xmath582 and the penultimate column of @xmath319 to obtain @xmath769 a fat rectangle and @xmath319 with @xmath785 ; so this case and the special cases follow by [ sec : rectangle ] .
@xmath1109 \draw[very thick ] ( 0,0)--(2.5,0)--(2.5,-1)--(0,-1)--(0,0 ) ; \draw ( 2.5,0)--(5,0)--(5,-0.5)--(4.5,-0.5)--(4.5,-1)--(2.5,-1 ) ; \draw(2.5,-1)--(2.5,-1.5)--(2,-1.5)--(2,-2)--(0,-2)--(0,-1 ) ; \end{tikzpicture}\end{minipage } \begin{minipage}{33mm}\begin{tikzpicture}[scale=0.55 ] \draw[very thick ] ( 0,1.5)--(2.5,1.5)--(2.5,-1)--(0,-1)--(0,1.5 ) ; \draw ( 2.5,1.5)--(3.5,1.5)--(3.5,-0.5)--(3,-0.5)--(3,-1)--(2.5,-1 ) ; \draw ( 2.5,-1)--(2.5,-1.5)--(2,-1.5)--(2,-2)--(0,-2)--(0,-1 ) ; \end{tikzpicture}\end{minipage } \begin{minipage}{30mm}\begin{tikzpicture}[scale=0.55 ] \draw[very thick ] ( 1.5,0)--(2.5,0)--(2.5,-1.5)--(1.5,-1.5)--(1.5,0 ) ; \draw ( 2.5,0)--(5,0)--(5,-0.5)--(4.5,-0.5)--(4.5,-1)--(2.5,-1 ) ; \draw(2.5,-1.5)--(2.5,-3.5)--(2,-3.5)--(2,-4)--(1.5,-4)--(1.5,-1 ) ; \end{tikzpicture}\end{minipage } \begin{minipage}{33mm}\begin{tikzpicture}[scale=0.55 ] \draw[very thick ] ( 1,0)--(2.5,0)--(2.5,-1)--(1.5,-1)--(1.5,-3.5)--(1,-3.5)--(1,0 ) ; \draw ( 2.5,0)--(5,0)--(5,-0.5)--(4.5,-0.5)--(4.5,-1)--(2.5,-1 ) ; \draw(2.5,-1.5+.5)--(2.5,-3.5+.5)--(2,-3.5+.5)--(2,-4+.5)--(1.5,-4+.5)--(1.5,-1 ) ; \end{tikzpicture}\end{minipage}\ ] ] continuing with case @xmath930 , suppose that @xmath67 is equal to one of @xmath455 , @xmath1110 or @xmath1111 and @xmath578 is a non - linear rectangle .
remove all rows and columns common to both @xmath135 and @xmath58 to obtain @xmath582 a proper fat hook and @xmath319 a non - linear rectangle ; the result follows by [ sec : rectangle ] .
it remains to consider the cases where @xmath67 is equal to one of @xmath1112 , @xmath669 or @xmath1113 or @xmath1114 and @xmath578 is a non - linear rectangle .
suppose @xmath1115 . if @xmath790 , remove all rows and columns common to @xmath58 and @xmath135 to obtain @xmath1116 and @xmath319 a fat rectangle .
if @xmath789 , remove all rows and columns common to @xmath58 and @xmath135 with the exception of one column in the leg to obtain @xmath1117 and @xmath1118 . remove the final two rows of @xmath582 and the final two columns of @xmath319 to obtain @xmath769 a fat rectangle and @xmath1119 a proper fat hook .
the result follows by [ sec : rectangle ] .
now suppose that @xmath1120 or @xmath1113 or @xmath1114 and that @xmath1121 ( as the other cases were handled above ) .
remove all rows and columns common to both @xmath135 and @xmath39 to obtain @xmath1122 a 2-line partition not of the form @xmath1123 or @xmath1124 and @xmath319 is a non - linear rectangle .
the result follows by [ section2parter ] .
now consider the cases where both @xmath67 and @xmath578 are equal to @xmath303 ( up to conjugation ) for @xmath1125 , where @xmath910 . remove all rows and columns common to both @xmath135 and @xmath39 and arrive at twelve distinct cases ( as @xmath932 and @xmath961 produce the same set of cases ) .
eleven of the twelve cases follow by [ squares , section2parter , sec : rectangle ] .
the final case is @xmath1126 . for @xmath838 , remove the final column of @xmath582 and the final row of @xmath319 to obtain a pair of rectangular partitions .
the result follows from [ sec : rectangle ] . for @xmath625
we obtain the seed @xmath1127 .
it remains to consider cases @xmath1125 in which precisely one of @xmath67 and @xmath578 is a rectangle . in case @xmath931 , where @xmath67 has to be a rectangle ( respectively in cases @xmath973 when @xmath578 is a rectangle ) remove all rows and columns common to @xmath135 and @xmath39 to obtain @xmath319 a non - linear rectangle and @xmath582 a proper fat hook ( respectively @xmath582 a rectangle and @xmath319 a proper fat hook ) .
the result follows from [ sec : rectangle ] .
it remains to consider cases @xmath973 for @xmath67 a rectangle . in cases
@xmath932 ( respectively @xmath961 ) with @xmath706 and @xmath1128 , remove all rows and columns common to @xmath58 and @xmath135 with the exception of one column in the leg ( respectively row in the arm ) to obtain @xmath319 and @xmath582 . in case @xmath961 , remove the final two columns of @xmath319 and the final two rows of @xmath582 to obtain @xmath319 a fat rectangle and @xmath582 a proper fat hook . in both cases the result follows by [ sec : rectangle ] . for a case not of the above form , remove all rows and columns common to @xmath58 and @xmath135 to obtain @xmath319 and @xmath582 .
the result follows by [ sec : rectangle ] . if either of @xmath67 or @xmath578 is linear and the other has two connected components , then @xmath312 . by [ theo : thdvir1,thm :
mf - outer , thm : mf - basic - skew ] , the non - linear diagram ( of the pair @xmath67 and @xmath578 ) belongs to the list of skew partitions in [ thm : mf - outer ] . by [ 42879342578837495278354278354 ] , we can assume that neither of @xmath775 $ ] or @xmath774 $ ] is @xmath1027 \boxtimes [ 1]$ ] , up to conjugation .
we first consider the exceptional cases where we can not remove all rows and columns common to @xmath135 and @xmath39 .
these only happen in a few cases in which all three external components in the diagram are linear .
up to conjugation of both @xmath58 and @xmath135 , our exceptional cases are listed below . by aggressive application of [ animportantremark ]
, we can remove all rows and columns common to both partitions with the exception of a single row , @xmath1129 , or column , @xmath1130 , to obtain @xmath828 .
these rows and columns are also listed . * @xmath67 linear , @xmath1131 , @xmath1132 and @xmath1130 a single column in the arm ; * @xmath586 linear , @xmath1131 , @xmath1133 and @xmath1129 a single row in the arm ; * @xmath1134 , @xmath1131 and @xmath1132 and @xmath1130 a single column in the leg ; * no exceptions ; * @xmath1135 , @xmath1136 and @xmath1130 a single column in the arm or the leg ; * @xmath1137 @xmath1138 and @xmath1136 or @xmath1139 and @xmath1129 a single row in the arm ( the resulting partitions are the same as in the @xmath1032 case ) . for any @xmath67 and @xmath578 and any case @xmath1140@xmath1141 not on the above list ,
remove all columns and rows common to @xmath135 and @xmath39 to obtain @xmath582 and @xmath319 .
for an example of how [ animportantremark ] is used , we compare @xmath1032 and @xmath1033 ; here we have reduced to the same set of exceptional cases , but using different arguments . for @xmath1032 , we know there must exist a row in the arm as @xmath135 is non - rectangular . for @xmath1033 , we know that @xmath58 is not a hook , and so there must be an extra row in the arm or column in the leg . however , case @xmath1033 is symmetric under conjugation ( note that case @xmath1032 is not ) and so we can assume there is an extra row in the arm . for the exceptional cases of type @xmath1030 , we have that @xmath582 is non - rectangular ( up to conjugation
, @xmath582 is obtained by adding a single node to the partition @xmath1142 ) and @xmath1143 is a hook partition .
therefore the result follows from [ sectionhook ! ] .
the generic case follows from [ sec : rectangle ] , as @xmath582 is a fat rectangle and @xmath1144 . in case @xmath1029 , we know by [ induction ] and [ thm : mf - outer ] that @xmath1145 is a fat hook and
therefore @xmath1146 is a rectangle . if @xmath1134 we remove the final @xmath1147 rows ( each of width @xmath1148 ) from @xmath582 and the final @xmath1149 rows ( each of width @xmath1150 ) from @xmath319 and hence obtain a pair @xmath1013 as in @xmath931 .
if @xmath1151 , then the shortest row of @xmath1152 ( equal to @xmath1153 ) is longer than the longest column of @xmath319 ( equal to @xmath1154 ) and therefore @xmath1155 and @xmath1156 are both connected ( in fact @xmath1157 are as in case @xmath930 ) . in both cases ,
the result follows by earlier results in this section .
we now consider case @xmath1045 . in the exceptional case with @xmath1130 a single column in the arm
, we have that @xmath1158 and @xmath1152 is a proper fat hook ; the result holds by [ short_second_row ] and [ induction ] . in the exceptional case with @xmath1130 a single column in the leg ,
we remove the final two columns of @xmath319 and the final row of @xmath582 to obtain @xmath1013 such that @xmath1159 and @xmath1160 are both proper skew partitions with two components each and the result follows by [ induction ] .
we now consider the generic case of @xmath1045 with @xmath1161 .
we first consider the case where @xmath1162 or @xmath1163 is equal to 1 .
if @xmath1164 and @xmath1165 is a rectangle , then @xmath319 is a hook and the result follows from [ sectionhook ! ] .
if @xmath1164 and @xmath1165 is not a rectangle , then remove @xmath1166 from @xmath582 and @xmath1167 nodes from @xmath1168 to obtain @xmath769 a fat rectangle and @xmath803 a partition with at least three removable nodes ; the result follows by [ sec : rectangle ] .
we now assume that @xmath1169 and @xmath1170 .
if @xmath1166 is a rectangle , then the result follows by [ section2parter ] .
if @xmath1166 is not a rectangle , remove the final @xmath1171 columns from @xmath582 and @xmath1172 nodes from @xmath1168 to obtain @xmath769 a non - linear rectangle and @xmath803 such that @xmath1173 . by [ animportantremark ]
( see [ thm : mf - outer ] in particular ) we may assume that at least one of @xmath1166 or @xmath1165 is a rectangle and that @xmath1174 . if @xmath1166 is a rectangle , remove @xmath1166 from the bottom of @xmath582 and @xmath1167 nodes from the @xmath1168 to obtain @xmath1175 a fat rectangle and @xmath1176 ; the result follows by [ sec : rectangle ] .
we now assume that @xmath1165 is a rectangle .
remove the final @xmath1171 columns of @xmath582 ( each of length @xmath1177 ) and @xmath1178 nodes from @xmath1168 to obtain @xmath1175 a non - linear rectangle and @xmath1179 .
the result follows by [ sec : rectangle ] .
we now consider the case @xmath1180 .
if @xmath1181 , @xmath1138 ( with @xmath1182 by our assumptions ) , remove the first row of @xmath582 and the final column of @xmath319 to obtain @xmath1183 ; the result follows by [ squares ] .
now assume that @xmath1180 and @xmath1184 are not of the above form .
the shortest column of @xmath1185 ( of length @xmath1186 ) is longer than the longest row of @xmath582 ( of length @xmath1187 ) and therefore @xmath1188 and @xmath1189 are both connected and the result follows by earlier results in this section .
we now consider the cases @xmath1033 and @xmath1032 .
we first consider the generic case of @xmath1033 .
if @xmath1166 and @xmath1165 are both rectangles , then @xmath321 are as in case @xmath1030 considered above .
if one of @xmath1166 and @xmath1165 is a rectangle and the other is a non - rectangular fat hook , then the pair @xmath321 are as in case @xmath1029 considered above . up to conjugation , it remains to consider the case in which @xmath1190 is linear and @xmath1191 is such that @xmath1192 ; in particular @xmath1193 .
if @xmath1161 , then the shortest row of @xmath319 is of length @xmath1194 , and the longest column of @xmath582 is less than or equal to @xmath1195 .
therefore @xmath1188 and @xmath1189 are both connected and the result follows by earlier results in this section . if @xmath1180 and @xmath1196 , remove the final @xmath1197 rows ( of width @xmath1198 ) from @xmath582 and the final @xmath1198 rows ( of width @xmath1199 ) from @xmath319 to obtain @xmath1013 . if @xmath1180 and @xmath1200 , remove the final @xmath1201 rows ( of width 1 ) from @xmath582 and the final @xmath1198 rows ( of width 2 ) from @xmath319 to obtain @xmath1013 . in either case , @xmath1202 and @xmath1203
are both connected and the result follows by earlier results in this section .
the generic case for @xmath1032 follows from [ sec : rectangle ] as @xmath582 is a rectangle .
we now argue for the exceptional case for @xmath1032 ( the exceptional case for @xmath1033 is identical but with the roles of @xmath67 and @xmath578 switched ) . for @xmath1134 ( respectively @xmath1204 )
remove the final row of @xmath582 ( respectively final two columns of @xmath582 ) and the final column of @xmath319 to obtain @xmath1205 such that @xmath1206 and @xmath1207 both having two connected components ( respectively @xmath321 are as in the generic case of @xmath1033 ) . for the exceptional case of @xmath1042 ,
remove the final row of @xmath582 and the final two columns of @xmath319 to obtain @xmath1013 .
if @xmath1208 , then @xmath1013 are as in the exceptional case for @xmath1032 . if @xmath1209 , then @xmath1159 has three connected components and so the result follows by [ induction ] .
now assume that we are in the generic case with @xmath940 .
if @xmath582 is a hook or 2-line partition , the result follows by [ sectionhook!,section2parter ] .
otherwise , we remove @xmath85 from @xmath319 and @xmath1210 nodes from the first column of @xmath582 to obtain a pair as in case @xmath961 with @xmath582 a proper fat hook . for @xmath934 , if @xmath1211 then the result follows by [ section2parter ] .
otherwise , @xmath1212 and the result follows by [ squares ] .
in this section , we continue to assume that [ thm : classification ] has been proven for all kronecker products labelled by pairs of partitions of degree less than or equal to @xmath38 .
armed with the proof of [ thm : classification ] for the case where one partition is a fat hook of degree @xmath25 , we now embark on proving the general case for arbitrary pairs of partitions of degree @xmath25
. we shall assume throughout that @xmath1213 are a pair of partitions such that @xmath373 and neither @xmath39 nor @xmath135 is a fat hook .
furthermore , by [ theo : thdvir1,animportantremark ] we may also assume that the pair of characters associated with the skew diagrams @xmath1214 and @xmath1215 belongs to the lists in [ thm : classification ] and [ thm : classification - skew ] . in particular , we may ( and will ) assume without loss of generality that @xmath67 has one connected component and that @xmath578 has either one or two connected components .
we shall systematically work through the list of possible pairs of shapes @xmath179 and @xmath178 and reduce the corresponding pairs of partitions @xmath58 and @xmath135 to pairs of partitions @xmath319
, @xmath582 such that @xmath1216 and the semigroup property implies @xmath312 .
our typical approach will be to reduce to the case that one of @xmath319 or @xmath582 is a 2-line , rectangle , or fat hook partition and then appeal to the results of [ sec : warm - up , sec : rectangle , sec : fatty ] .
we let @xmath1218 , @xmath1219 and we suppose , without loss of generality , that @xmath1220 and @xmath1221 .
our general strategy shall be to remove all rows and columns outside of the region labelled by @xmath1222 \times [ c_2 , \dots , c_1]$ ] , an example is depicted in [ asdfasfsdafas ] , below .
we first consider the exceptional cases in which @xmath1223 \times [ c_2 , \dots , c_1 ] ) \cap \lambda \cap \mu\ ] ] is equal to the young diagram of a partition of the form @xmath15 , @xmath1106 , @xmath616 , or @xmath615 up to conjugation .
@xmath1224 \draw[very thick ] ( -1,1.5)--(5,1.5)--(5,1)--(4,1)--(4,0.5)--(3.5,0.5)--(3.5,0)--(2.5,0 ) --(2.5,-1)--(2,-1)--(2,-2)--(1.5,-2)--(1.5,-2.5)--(0.5,-2.5)-- ( 0.5,-3)-- ( 0.5,-3.5)-- ( -1,-3.5)-- ( -1,1.5 ) ; \draw[thick , dotted ] ( -2,-2)--(5,-2 ) ( -2,0.5)--(5,0.5 ) ( 3.5,2.5)--(3.5,-3.5 ) ( 1.5,2.5)--(1.5,-3.5 ) ; \draw(-2,-1.75 ) node[left ] { $ r_2 $ } ; \draw(-2,0.25 ) node[left ] { $ r_1 $ } ; \draw(1.75,2.5 ) node[above ] { $ c_2 $ } ; \draw(3.25,2.5 ) node[above ] { $ c_1 $ } ; \draw(1.75,-1.75 ) node { $ \times$ } ; \draw(3.25,0.25 ) node { $ \times$ } ; \end{tikzpicture}\end{minipage } \quad\quad\quad \begin{minipage}{38mm}\begin{tikzpicture}[scale=0.55 ] \draw[very thick ] ( -1,1.5)--(5,1.5)--(5,1)--(4,1)--(4,0.5)--(3.5,0.5)--(3.5,0)--(2.5,0 ) --(2.5,-1)--(2,-1)--(2,-2)--(1.5,-2)--(1.5,-2.5)--(0.5,-2.5)-- ( 0.5,-3)-- ( 0.5,-3.5)-- ( -1,-3.5)-- ( -1,1.5 ) ; \draw[thick , dotted ] ( -2,-2)--(5,-2 ) ( -2,-0.5)--(5,-0.5 ) ( 2.5,2.5)--(2.5,-3.5 ) ( 1.5,2.5)--(1.5,-3.5 ) ; \draw(-2,-1.75 ) node[left ] { $ r_2 $ } ; \draw(-2,-0.75 ) node[left ] { $ r_1 $ } ; \draw(1.75,2.5 ) node[above ] { $ c_2 $ } ; \draw(2.35,2.5 ) node[above ] { $ c_1 $ } ; \draw(1.75,-1.75 ) node { $ \times$ } ; \draw(2.25,-0.75 ) node { $ \times$ } ; \end{tikzpicture}\end{minipage}\ ] ] we may assume that @xmath1225 ( the case @xmath1226 is similar ) ; we remove most rows common to @xmath58 and @xmath135 to obtain three distinct cases . if both @xmath1227 , we can remove all but one column to the left of the region and all but one row above the region from @xmath145 to obtain partitions @xmath828 such that @xmath1228 in this case @xmath1229 , by [ sec : fatty ] .
now suppose that @xmath1230 , i.e. , @xmath1231 . by assumption , we have that @xmath59 is not a rectangle and
so @xmath1232 . we remove all but the longest row ( of width @xmath1233 ) above the region ; we then truncate this row to be of length @xmath1234 ; we hence obtain @xmath319 and @xmath582 such that @xmath1235 in this case @xmath1236 , by [ section2parter ] .
now suppose that @xmath1237 , in which case we can remove all but the longest column to the left of the region ( which is of length greater than or equal to 3 , by assumption that neither of @xmath58 or @xmath135 is a 2-line partition ) , we then truncate this column to be of length 3 and hence obtain @xmath319 and @xmath582 such that @xmath1238 in this case @xmath1239 , by [ sectionhook ! ] .
we now assume that @xmath1240 is of the form @xmath15 , @xmath303 , or @xmath616 up to conjugation . in all of these cases , we know that there is at least one extra column or row common to @xmath58 and @xmath135 which we may consider ; this follows from our assumption that neither @xmath58 nor @xmath135 is a 2-line partition .
this leads us to define @xmath828 as the intersections of @xmath145 with the region @xmath1241 \times [ c_2 , \dots , c_1]$ ] or @xmath1222 \times [ c_2 - 1,c_2 , \dots , c_1]$ ] , so that @xmath1242 is equal to one of @xmath1243 \times [ c_2 , \dots , c_1 ] ) \cap \lambda \cap \mu \qquad \text{or } \qquad ( [ r_1 , \dots , r_2 ] \times [ c_2 - 1,c_2 , \dots , c_1 ] ) \cap \lambda \cap \mu\:.\ ] ] it will then suffice to show that @xmath357 in both cases for all three possible partitions , @xmath1244 . in the latter case , for @xmath1245 we have that @xmath1246 ; removing @xmath435 from @xmath1168 and @xmath435 from @xmath1247 , the result follows from [ section2parter ] . in the other five cases , the result follows as @xmath321 is not on the list of [ thm : classification ] and one of the two partitions is a fat hook and so the result follows by [ sec : fatty ] .
we now deal with the generic case ( in which @xmath1248 , @xmath15 , @xmath303 , or @xmath616 , up to conjugation ) ; remove all rows and columns outside of the region labelled by @xmath1222 \times [ c_2 , \dots , c_1]$ ] from @xmath145 , to obtain @xmath319 and @xmath582 such that @xmath1249 \times [ c_2 , \dots , c_1 ] \cap \lambda \cap \mu.\ ] ] we note that the node @xmath67 ( respectively @xmath1250 is @xmath132-removable from @xmath582 ( respectively @xmath319 ) ; the result follows as @xmath1251 by [ selftensor ] . we assume , without loss of generality , that @xmath67 appears higher than @xmath578 in the diagram and @xmath1252 , @xmath1253 or @xmath1254 . by [ firstlinear ]
, we may assume that @xmath34 .
the case @xmath1255 can be obtained by conjugation .
* case 1 : * @xmath1256 .
assume there is a column , @xmath1130 , to the left of @xmath578 ( respectively to the right of @xmath67 ) .
remove from the intersection all rows and columns excluding column @xmath1130 ( respectively all columns excluding @xmath1130 , and all but one of the rows of width @xmath1257 above @xmath67 ) to obtain @xmath321 equal to either of @xmath1258 and the result follows from [ sec : fatty ] .
now assume there is no such column to the left or right and recall our assumption that neither @xmath39 nor @xmath135 is a 2-line partition .
there are two distinct cases to consider , namely * @xmath34 and there is a single column , @xmath1130 , in between @xmath578 and @xmath67 and a row , @xmath1129 , above @xmath67 ; * @xmath170 and there are at least two columns , @xmath1259 in between @xmath67 and @xmath578 and no rows above @xmath67 . in the former case ,
we remove from the intersection all rows and columns excluding @xmath1129 and @xmath1130 to obtain @xmath1260 . in the latter case ,
we remove from the intersection all rows and all columns except @xmath1261 and @xmath1262 to obtain @xmath1263 . in both cases the result follows from [ sec : rectangle ] as
@xmath582 is a rectangle .
* case 2 : * @xmath1264 for @xmath34 .
if there is both a column and a row between @xmath67 and @xmath578 , then we reduce to the case @xmath1265 and the result follows [ section2parter ] .
we may now assume that there is not both a column and a row between @xmath67 and @xmath578 .
conjugating if necessary , we may assume that there is no column between @xmath67 and @xmath578 .
suppose that there are no rows above @xmath67 . then by our assumption that @xmath1266 , there are two columns @xmath1130 and @xmath1267 to the left of @xmath578 .
we remove from the intersection all rows and columns except for @xmath1130 and @xmath1267 to obtain @xmath1268 and the result follows from [ sec : rectangle ] .
we may now suppose that there is a row , @xmath1129 , above @xmath67 . by assumption ,
@xmath319 is not a hook partition and so there is either @xmath195 a single column , @xmath1130 , to the left of @xmath578 or @xmath196 an extra row @xmath1269 above @xmath67 . in the former case , we remove all rows except @xmath1129 and all columns except @xmath1130 and hence obtain @xmath1270 with @xmath34 and so the result follows from [ sec : fatty ] . in the latter case , we remove from the intersection all rows and columns with the exception of @xmath1129 and @xmath1269 to obtain @xmath1271 ; the result follows from [ sec : rectangle ] .
* case 3 : * @xmath1272 . for @xmath625
, we can remove all but one column to the left of @xmath578 or all but one column between @xmath67 and @xmath578 ( up to conjugation ) to obtain @xmath321 equal to either of the small seeds @xmath1273 .
otherwise , we may remove all rows and columns common to both partitions , and the result follows by [ selftensor ] .
we assume , without loss of generality , that @xmath586 is linear and that it appears higher than @xmath578 in the diagram . by [ 8.2 ] , we can assume that @xmath578 is non - linear .
we start with the discussion of the cases where @xmath578 is a proper partition .
* case 1 : * @xmath934 and @xmath578 is a proper partition .
suppose there are no rows either above @xmath67 or between @xmath67 and @xmath578 .
in which case ( by our assumption that @xmath135 is neither linear , nor a hook ) there exist two columns @xmath1130 and @xmath1267 to the left of @xmath578 .
we remove from the intersection all rows and all columns with the exception of @xmath1130 and @xmath1267 .
the result follows as @xmath582 is a fat hook .
suppose that there is a row , @xmath1129 , above @xmath67 .
remove all rows and columns common to both @xmath58 and @xmath135 with the exception of @xmath1129 , to obtain @xmath1274 and @xmath1275 ; we have that @xmath319 is either a proper fat hook or @xmath980 ( our assumptions imply that @xmath1276 , @xmath170 ) .
the result then follows from [ section2parter ] .
now assume that there is a row , @xmath1129 , between @xmath67 and @xmath578 and no row above @xmath67 .
if @xmath578 is not a fat hook , then remove all common rows and columns from @xmath58 and @xmath135 with the exception of @xmath1129 to obtain @xmath582 a 2-line partition and @xmath319 a partition such that @xmath980 .
we now assume that @xmath578 is a fat hook . now
( by assumption that @xmath135 is not a 2-line partition ) there is either a second row @xmath1269 between @xmath67 and @xmath578 or an extra column , @xmath1130 , to the left of @xmath578 . in either case
remove all rows and columns from @xmath58 and @xmath135 with the exception of @xmath1129 and @xmath1269 or @xmath1129 and @xmath1130 to obtain a pair @xmath321 . in the former case
, @xmath582 is a proper fat hook and @xmath1277 $ ] is neither a linear character nor the natural character or its dual . in the latter case ,
@xmath319 is a proper fat hook and @xmath582 has three removable nodes . in either case , the result follows from [ sec : fatty ] .
* case 2 : * @xmath940 and @xmath578 is a proper partition .
we have two exceptional cases to consider , in which @xmath1278 or @xmath1279 . in either case
, we remove all but a single row or column from @xmath58 and @xmath135 to obtain 12 seeds @xmath321 of degree less than or equal to 18 .
assume @xmath1280 , or @xmath1279 ; remove all rows and columns common to @xmath58 and @xmath135 to obtain @xmath319 and @xmath582 .
if @xmath1281 , the result follows from [ section2parter ] . otherwise , @xmath1282 with @xmath1283 and so @xmath1284 by [ squares ] . * case 3 : * @xmath934 and @xmath1073 is a proper partition . by assumption ,
@xmath58 is not a fat hook and so there exists at least two columns , @xmath1130 and @xmath1267 , of _ distinct lengths _ belonging to one or two of the regions : to the left of @xmath578 , between @xmath67 and @xmath578 , or to the right of @xmath67 .
we can assume that the final node in column , @xmath1130 say , does not belong to the same row as the nodes in the partition @xmath67 .
remove all rows and columns except for @xmath1130 to obtain @xmath319 a proper fat hook and @xmath582 such that @xmath1285 ; the result follows from [ sec : fatty ] .
* case 4 : * @xmath940 and @xmath1073 is a proper partition .
we remove all rows and columns common to @xmath58 and @xmath135 , to obtain @xmath319 a non - linear rectangle and @xmath582 a non - rectangular partition such that @xmath1286 ; the result follows from [ sec : rectangle ] .
we fix some notation which will be used throughout the remainder of this section .
if @xmath578 and @xmath67 each have exactly one connected component , then we can assume without loss of generality that @xmath578 lies below @xmath67 on the diagram , as depicted in the leftmost diagram in [ extra rows and columns ] .
we shall let @xmath1287 ( respectively @xmath1288 ) denote the longest row in @xmath1289 which appears above @xmath67 ( respectively between @xmath578 and @xmath67 ) if such a row exists , and let @xmath1287 ( respectively @xmath1288 ) be undefined otherwise .
similarly , we shall let @xmath1261 ( respectively @xmath1262 ) denote the longest column in @xmath1289 which appears to the left of @xmath578 ( respectively between @xmath578 and @xmath67 ) if such a column exists , and let @xmath1261 ( respectively @xmath1262 ) be undefined otherwise .
this is depicted in [ extra rows and columns ] .
* @xmath67 lies below @xmath85 and @xmath86 on the diagram , as depicted in the middle diagram in [ extra rows and columns ] ; * @xmath67 lies between @xmath85 and @xmath86 on the diagram , as depicted in the rightmost diagram in [ extra rows and columns ] .
@xmath1292 \draw[very thick ] ( -2,1.5)--(4.5,1.5)--(4.5,1)--(4,1)--(4,0.5)--(3.5,0.5)--(3.5,0)--(2.5,0 ) --(2.5,-1)--(2,-1)--(2,-2)--(1.5,-2)--(1.5,-2.5)--(0.5,-2.5)--(0.5,-4)--(0,-4)--(0,-4.5)--(-0.5,-4.5)--(-0.5,-5 ) --(-2,-5)--(-2,1.5 ) ; \draw[very thick ] ( 3.5,0.5)--(3.5,-1.5)--(2,-1.5 ) ; \draw[very thick ] ( 1.5,-2.5)--(-0.5,-2.5)--(-0.5,-4.5 ) ; \draw(0,-3.25 ) node { $ \gamma$ } ; \draw(3,-0.8 ) node { $ \delta$ } ; \draw(-2.5,1.2 ) node { $ c_1 $ } ; \draw(-1.75,1.25)--(4.25,1.25 ) ; \draw(-2.5,-1.75 ) node { $ c_2 $ } ; \draw(-1.75,-1.75)--(1.75,-1.75 ) ; \draw(-1.65,1.9 ) node { $ r_1 $ } ; \draw(-1.75,1.25)--(-1.75,-4.75 ) ; \draw(1.25,1.9 ) node { $ r_2 $ } ; \draw(1.25,-2.25)--(1.25,1.25 ) ; \end{tikzpicture}\end{minipage } \quad \begin{minipage}{48mm}\begin{tikzpicture}[scale=0.5 ] \draw[very thick ] ( -2,3.5)--(6,3.5)--(6,3.5)--(6,3)--(5.5,3)--(5.5,2.5)--(4.5,2.5)--(4.5,1)--(4,1)--(4,0.5)--(3.5,0.5)--(3.5,0)--(2.5,0 ) --(2.5,-1)--(2,-1)--(2,-2)--(1.5,-2)--(1.5,-2.5)--(0.5,-2.5)--(0.5,-4)--(0,-4)--(0,-4.5)--(-0.5,-4.5)--(-0.5,-5 ) --(-2,-5)--(-2,3.5 ) ; \draw[very thick ] ( 3.5,0.5)--(3.5,-1.5)--(2,-1.5 ) ; \draw[very thick ] ( 1.5,-2.5)--(-1,-2.5)--(-1,-5 ) ; \draw(-.25,-3.25 ) node { $ \gamma$ } ; \draw(3,-0.8 ) node { $ \delta''$ } ; \draw[very thick ] ( 5.5,3)--(5.5,1.5)--(4.5,1.5 ) ; \draw(5,2 ) node { $ \delta'$ } ; \draw(-2.5,1.2 ) node { $ r_2 $ } ; \draw(-1.75,1.25)--(4.25,1.25 ) ; \draw(-2.5,-1.75 ) node { $ r_3 $ } ; \draw(-1.75,-1.75)--(1.75,-1.75 ) ; \draw(-1.65,4 ) node { $ c_1 $ } ; \draw(-1.75,3.25)--(-1.75,-4.75 ) ; \draw(1.25,4 ) node { $ c_2 $ } ; \draw(1.25,-2.25)--(1.25,3.25 ) ; \draw(-2.5,3.25 ) node { $ r_1 $ } ; \draw(-1.75,3.25)--(5.75,3.25 ) ; \draw(3.75,4 ) node { $ c_3 $ } ; \draw(3.75,0.75)--(3.75,3.25 ) ; \end{tikzpicture}\end{minipage } \quad \begin{minipage}{38mm}\begin{tikzpicture}[scale=0.5 ] \draw[very thick ] ( -2,3.5)--(6,3.5)--(6,3.5)--(6,3)--(5,3)--(5,2.5)--(4.5,2.5)--(4.5,1)--(4,1)--(4,0.5)--(3,0.5)--(3,0.5)--(2.5,0.5 ) --(2.5,-0.5)--(2,-0.5)--(2,-2)--(1.5,-2)--(1.5,-2.5)--(0.5,-2.5)--(0.5,-4)--(0,-4)--(0,-4.5)--(-0.5,-4.5)--(-0.5,-5 ) --(-2,-5)--(-2,3.5 ) ; \draw[very thick ] ( 3.5,0.5)--(3.5,-1.5)--(2,-1.5 ) ; \draw[very thick ] ( 1.5,-2.5)--(0.5,-2.5)--(0.5,-3)--(-0.5,-3)--(-0.5,-4.5 ) ; \draw(0,-3.5 ) node { $ \delta '' $ } ; \draw(3,-0.8 ) node { $ \gamma$ } ; \draw[very thick ] ( 5.5,3)--(5.5,1.5)--(4.5,1.5 ) ; \draw(5,2 ) node { $ \delta'$ } ; \draw(-2.5,1.2 ) node { $ r_2 $ } ; \draw(-1.75,1.25)--(4.25,1.25 ) ; \draw(-2.5,-1.75 ) node { $ r_3 $ } ; \draw(-1.75,-1.75)--(1.75,-1.75 ) ; \draw(-1.65,4 ) node { $ c_1 $ } ; \draw(-1.75,3.25)--(-1.75,-4.75 ) ; \draw(1.25,4 ) node { $ c_2 $ } ; \draw(1.25,-2.25)--(1.25,3.25 ) ; \draw(-2.5,3.25 ) node { $ r_1 $ } ; \draw(-1.75,3.25)--(5.75,3.25 ) ; \draw(3.75,4 ) node { $ c_3 $ } ; \draw(3.75,0.75)--(3.75,3.25 ) ; we assume , without loss of generality , that @xmath67 appears higher than @xmath578 in the diagram . by [ thm : mf - basic - skew ] and [ induction ] , we know that @xmath578 is of the form @xmath1293 where @xmath80 is a partition , and @xmath81 is a rectangle , or @xmath1294 satisfy @xmath1295 and @xmath1296 .
we consider the exceptional case in which @xmath1297 and @xmath1298 and @xmath934 . by assumption ,
neither @xmath58 nor @xmath135 is a fat hook and so there exists at least one extra row or column @xmath1299 or @xmath1262 as in [ extra rows and columns ] .
we remove all rows and columns common to both @xmath58 and @xmath135 with the exception of one of @xmath1287 or @xmath1288 or @xmath1261 or @xmath1262 to obtain @xmath1300 . in all other cases ,
we remove all rows and columns common to both @xmath58 and @xmath135 to obtain a pair of partition @xmath325 .
the resulting pair @xmath321 are such that @xmath195 @xmath1301 @xmath196 both @xmath319 and @xmath582 are non - rectangular @xmath213 neither @xmath1277,[{\tilde \mu}]$ ] is equal to the standard character or its dual .
therefore @xmath357 as required .
suppose that @xmath86 is below @xmath67 and @xmath67 is below @xmath85 as depicted in the rightmost diagram in [ extra rows and columns ] .
we assume without loss of generality that @xmath1302 and @xmath85 and @xmath86 are partitions of @xmath1303 and @xmath1304 , respectively .
the only exceptional case for such a shape is given by @xmath1305 and @xmath1306 .
we want to remove all but a single row or column from @xmath58 and @xmath135 depending on having a suitable row or column in one of the six cases illustrated in [ extra rows and columns ] ; however , as we assume that @xmath135 is not a 2-row partition , we can ignore the two cases @xmath1262 and @xmath1307 .
it therefore remains to consider the cases where one of the columns or rows @xmath1308 and @xmath1309 exist , and we have reduced all other rows and columns common to both @xmath58 and @xmath135 to obtain @xmath319 , @xmath582 . in each of these four cases ,
the partition @xmath582 is either a proper fat hook or a fat rectangle and @xmath319 is a partition with @xmath1310 , @xmath1311 and @xmath1312 ; the assertion follows from the result for fat hooks .
suppose that @xmath67 is below @xmath1313 , and @xmath86 is below @xmath85 as depicted in the central diagram in [ extra rows and columns ] . the only exceptional case for such a shape is given by @xmath1314 , @xmath1305 and @xmath1306 . in this case ,
we need to consider each of the six possible cases given by removing all rows and columns common to @xmath58 and @xmath135 with the exception of one of @xmath1315 or @xmath1307 to hence obtain partititons @xmath319 and @xmath582 . in the case of @xmath1287 , @xmath1309 , @xmath1261 or @xmath1262
, we have that one of the partitions @xmath1316 is a fat hook and the other has 3 removable nodes . in the case of @xmath1288 or @xmath1307 , we have that @xmath1317 and either @xmath319 or @xmath582 has width and length at least 3
. therefore the claim follows from the result for fat hooks .
having taken care of the exceptional cases , we now turn our attention to the generic case . by our inductive assumption ,
we have that one of @xmath85 and @xmath86 is a rectangle and the other is a proper partition , up to rotation .
note that this covers all the pairs @xmath85 and @xmath86 in [ thm : classification - skew , thm : mf - outer ] .
we let @xmath319 and @xmath582 denote the partitions obtained by removing all row and columns common to both @xmath58 and @xmath135 .
we first cover the simplest case in which @xmath85 and @xmath86 are both rectangles ( and one may be linear ) . in this case
, we remove all rows and columns common to @xmath58 and @xmath135 to obtain a pair of partitions @xmath1301 which are both fat hooks and do not give a pair on our list ; the result follows .
we now assume that one of @xmath85 and @xmath86 is a rectangle and the other is a proper non - rectangular partition up to rotation .
if one of @xmath85 and @xmath86 is a rectangle and the other is obtained by rotating a proper non - rectangular partition , then @xmath319 is necessarily a proper fat hook and @xmath582 is either a proper fat hook or @xmath1318 , and the result follows . in the non - rotated case
, @xmath582 is necessarily a fat hook and @xmath1319 , and the result follows . by [ thm : classification - skew ] and [ 8.4,finallinear ] , it only remains to check the case where @xmath1320 , for @xmath248 , and @xmath774=[\delta']\boxtimes [ \delta'']$ ] with one of @xmath85 , @xmath86 being @xmath757 and the other linear .
we remove all rows and columns common to both @xmath58 and @xmath135 to obtain a pair or partitions @xmath321 .
we can assume without loss of generality that @xmath67 appears below @xmath85 and @xmath86 or between @xmath85 and @xmath86 .
in the former case , @xmath582 is a proper fat hook and @xmath1321 ; the result follows . in the latter case
@xmath582 is a fat rectangle and @xmath319 is a partition satisfying @xmath1322 and @xmath1323 ; therefore the result holds .
given the previous results , we suppose without loss of generality that @xmath67 is a non - linear rectangle and @xmath578 is a non - linear fat hook up to rotation .
we assume without loss of generality that @xmath67 appears above @xmath578 , as in [ extra rows and columns ] .
there are numerous exceptional small cases , however we do not need to list them all . instead , we shall show that if there is a row or column @xmath1324 , or @xmath1262 as in [ extra rows and columns ] then the product contains multiplicities .
if there is no such row or column in the diagram for @xmath58 and @xmath135 then if @xmath578 ( respectively @xmath1073 ) is a proper partition , then it follows that @xmath135 ( respectively @xmath58 ) is a rectangular partition ( recall that @xmath145 are not 2-line partitions , so the pair @xmath1325 does not occur ) , and so we are done .
now suppose that the diagram has a row or column @xmath1324 , or @xmath1262 and we let @xmath321 denote the pair obtained by removing all common rows and columns except this single row or column ( in each of the four cases ) ; we now show that the product contains multiplicities .
if there is a row , @xmath1287 , in the diagram for @xmath58 and @xmath135 , then @xmath582 is either a proper fat hook or a fat rectangle and @xmath319 is such that @xmath1326 . if there is a column , @xmath1261 , then @xmath582 is either a proper fat hook or @xmath1327 , and @xmath319 is either a proper fat hook or a fat rectangle .
if there is a row , @xmath1288 , and @xmath578 is a proper partition ( respectively @xmath1073 is a proper partition and @xmath578 is not ) , then @xmath582 is a proper fat hook ( respectively @xmath1327 ) and @xmath1328 ( respectively @xmath319 is a rectangular partition ) .
if there is a column , @xmath1262 , and @xmath578 is a proper partition ( respectively @xmath1073 is a proper partition and @xmath578 is not ) then @xmath582 is a non - linear rectangle ( respectively is a proper fat hook ) and in either case @xmath319 is either a proper fat hook or @xmath980 . in each of these cases ,
the result follows by the result for rectangles and fat hooks . in case @xmath195
, we remove all rows and columns to obtain @xmath319 and @xmath582 a pair of proper fat hooks and the result follows . in case @xmath196
, we remove all rows and columns to obtain @xmath319 a rectangular partition and @xmath582 such that @xmath1327 , the result follows . in case @xmath213 , we first deal with the exceptional case , in which @xmath1329 . we remove all but one row or column @xmath1330 to obtain a pair @xmath321 . in the case of @xmath1261 ( respectively @xmath1287 ) the partition @xmath319
( respectively @xmath1331 is a proper fat hook and @xmath582 ( respectively @xmath319 ) has 3 removable nodes , the result follows . in the case of @xmath1262 ( respectively @xmath1288 ) the partition @xmath319 ( respectively @xmath1331 is a 2-line partition and @xmath582 ( respectively @xmath319 ) has 3 removable nodes , the result follows .
now suppose @xmath1332 and @xmath1281 , in which case @xmath1333 and @xmath1334 ; the result follows from the case for 2-line partitions . if @xmath1335 , then @xmath321 are a pair of proper fat hooks and the result follows . in all three cases ,
let @xmath321 denote the pair of partitions obtained by removing all rows and columns common to both @xmath58 and @xmath135 . in case @xmath195 , we have that @xmath319 is a rectangular partition and @xmath582 is such that @xmath980 . in case
@xmath196 we have that @xmath321 is a pair of proper fat hooks . in case @xmath213
, we have that @xmath1336 and the pair is not on the list of [ thm : classification ] .
the result follows .
* acknowledgements .
* we would like to thank the american institute of mathematics and the organisers of `` combinatorics and complexity of kronecker coefficients '' and banff international research station for their hospitality .
we would also like to thank leibniz university hannover , epsrc grant ep / l01078x/1 , and the royal commission for the exhibition of 1851 for their financial support towards mutual visits in hannover and london .
we would like to thank s. assaf , a. hicks , j. remmel , v. tewari , and s. van willigenburg for their interest at some stages of this project .
finally , we would like to thank m. wildon for asking us about [ thm:3classification ] during the conference `` kronecker coefficients 2016 '' .
k. mulmuley , geometric complexity theory vi : the flip via saturated and positive integer programming in representation theory and algebraic geometry , tech .
report tr2007 - 04 , computer science department , the university of chicago , may 2007 . | we provide a classification of multiplicity - free inner tensor products of irreducible characters of symmetric groups , thus confirming a conjecture of bessenrodt .
concurrently , we classify all multiplicity - free inner tensor products of skew characters of the symmetric groups . we also provide formulae for calculating the decomposition of these tensor products . |
eclipse mapping is an inversion technique that can be used to reconstruct the distribution of bright material in the accretion streams of eclipsing polars .
the eclipsing nature of these systems means that for a short duration the stream is the dominant contributor to the observed brightness , the white dwarf and accretion region being hidden from view by the secondary .
the technique of eclipse mapping was first applied to accretion disks in the early 1980s , and the method was later applied to the bright eclipsing polar hu aqr by hakala ( 1995 ) . because it was found that the accretion stream contributed around half the optical emission in hu aqr , this system was ideal for the application of eclipse mapping to determine the distribution of this bright stream material in the binary system .
hakala ( 1995 ) developed this method assuming a one - dimensional stream confined to the orbital plane .
this was represented as an arc connecting the inner lagrangian point ( @xmath0 ) to the white dwarf .
developments to this method included a two part trajectory , and models using this have been successfully applied to the eclipse light curves of the bright eclipsing polars hu aqr and uz for ( bridge et al .
2002 ; vrielmann & schwope 2001 ; harrop - allin , potter & cropper 2001 ; kube , gnsicke & beuermann 2000 ; harrop - allin et al .
this trajectory comprises a ballistic free - fall component from the @xmath0-point ( lubow & shu 1975 ) , followed at some distance from the white dwarf by a magnetically confined component .
the confined part of the trajectory originates in the threading region from where the stream material is channeled by the field lines to the white dwarf surface .
the field lines are assumed to have a dipolar geometry and be centred on the white dwarf .
however , such a clear division between two trajectories is likely to be an inadequate assumption ( e.g. heerlein , horne & schwope 1999 ) and the assumption of a ballistic free - fall component may not be appropriate in all cases ( e.g. the low state of hu aqr ; harrop - allin et al .
a further development is a move to three - dimensional streams ( vrielmann & schwope 2001 ; kube et al .
2000 ) , thus attempting to include the effects of uneven heating of the stream and features related to the eclipse by the stream of the white dwarf and accretion region .
all these previous model techniques have the common feature that the stream trajectory is fixed prior to the application of the model .
hakala , cropper & ramsay ( 2002 ) introduced a technique which makes no assumptions about the location of the stream material prior to the modeling process . in principle , emission is allowed from anywhere within the roche lobe of the white dwarf .
they have successfully applied this technique to synthetic data sets , but here we apply it to observed light curves for the first time .
we model the ` white ' light curves of the eclipsing polars ep dra and hu aqr ( see bridge et al .
2002 , 2003 for a discussion of the light curves ) , observations of which were obtained with the superconducting tunnel junction ( stj ) camera s - cam 2 ( see e.g. perryman et al .
2001 ) . the s - cam 2 camera records the location on the array , time of arrival and energy of each incident photon .
this allows the study of the energy dependence of short timescale variations observed in polars .
we discuss briefly here the main points of the model ( for a detailed discussion see hakala et al .
the model consists of a number of ` fire - flies ' , where each fly is an individual bright emission point .
a fly is created with an initially random location within the roche lobe of the white dwarf , but is subsequently free to move during the evolution of the model .
the fly has a brightness ( @xmath1 ) which varies with the angle @xmath2 between the observer , the fly and the white dwarf .
this is defined as @xmath3 where @xmath4 is the minimum brightness of a fly and @xmath5 is the amplitude of angular dependence .
the brightness of those flies not eclipsed by the secondary at each phase are summed to form the model light curve .
many swarms of such flies are created and for each a model light curve can be generated .
the goodness of fit of an individual model light curve to the observed light curve is then defined using a fitness function , @xmath6 , which consists of a @xmath7 term and a regularisation term : @xmath8 .
the best fit solution is ` bred ' through a number of generations using a genetic algorithm ( ga ; see e.g. charbonneau 1995 for a review ) that is regulated by a self - organising map ( som ; kohonen 1990 ) .
the regularisation places a number of sections of a smooth curve through the fly distribution ( see hakala et al .
2002 ) , this helps to constrain the model where generally there are more degrees of freedom than data points .
the two end points of the curve are defined as the @xmath0-point and the white dwarf surface , thus providing two physically realistic constraints on the stream trajectory .
the model then prefers those swarms with minimal distance to the curve , resulting in smooth tube - like fly distributions .
hakala et al .
( 2002 ) also demonstrated the use of the technique with no regularisation term .
here there is no computational pressure on the flies to create any form of tube and they are not constrained to lie between the white dwarf and the @xmath0-point . in this case the model will define those regions of the roche lobe volume where emission can and can not be .
the light curves were obtained in october 2000 at the william herschel telescope , la palma .
preliminary results for fits to the ` white ' light curves of ep dra and hu aqr are shown in figure [ fig : fig01 ] , the light curves are binned into 3 s time intervals , with the model fits superimposed as a solid line .
we used 500 swarms each with 200 flies , and the models are run for 500 generations .
this provides a sufficient number of swarms to produce a unique solution and enough iterations to reach convergence .
figures [ fig : fig02 ] and [ fig : fig03 ] show the corresponding fly distributions .
the fitting process is necessarily a trade - off , mediated by the value of @xmath9 , between fitting real gradient changes and avoiding statistical noise features .
the presence of noise in the data will have the effect of broadening the fly distribution , but we can attempt to minimise this by adjusting the @xmath9 term to place more emphasis on the regularisation . however , while increasing @xmath9 has the effect of smoothing the distribution of flies into a more tube - like structure , we must be careful not to place too much emphasis on @xmath9 as this will have the effect of smoothing out any real brightness variations .
figure [ fig : fig02 ] shows that the emission points in hu aqr are located mostly out of the orbital plane .
the extent in volume of the emission points , and location at different distances in the z - direction may indicate that material is threaded along many different field lines , and may be the cause of the lack of evidence for a bright threading region at any one place in the orbital plane .
in contrast to hu aqr , the emission points in ep dra are located close to the orbital plane , possibly indicating accretion along field lines that meet the white dwarf surface at low latitudes . the emission in ep dra
is also distributed over a larger volume than hu aqr , particularly the region where we expect coupling to field lines to occur . to some extent
this is a product of a noisier light curve , but it could also indicate material located in a large stagnation region , where it collects before coupling to the white dwarf field lines .
we have applied this more objective technique to real data in an attempt to make the least number of assumptions about the location of bright material in these eclipsing binary systems .
we find that previous assumptions are largely consistent with our results .
for both ep dra and hu aqr most of the emission is located in the region where we expect material to be confined to the magnetic field lines of the white dwarf . in particular both models
show an enhanced concentration of emission points towards the white dwarf
. this may be indicative of where stream material is heated by x - ray emission from the accretion region on the white dwarf .
however , the contribution from the accretion region , seen as the large amplitude change at ingress and egress , is fixed for the duration of the model .
therefore , if the input value is too low the model will assign flies to the white dwarf to boost the contribution at the ingress and egress of the accretion region
. the number of flies near to the white dwarf may be indicative of this effect , rather than a real stream brightness enhancement .
an important advantage of placing no constraint on the stream trajectory , is the lack of sensitivity to input parameters .
the hu aqr data presented here has been modeled previously with a version of the technique of harrop - allin et al .
( 1999a ) .
this method was found to be acutely sensitive to the exact value chosen for the distance to the threading region , and also the location of the accretion region on the white dwarf surface , which fixes the geometry of the magnetic field lines ( bridge et al .
this meant that the application to the data was restricted , and results had to be interpreted accordingly .
as this new technique does not rely on the choice of initial parameters for the stream trajectory , we reduce these uncertainties . the stream trajectory used previously by bridge et al .
( 2002 ) to model the hu aqr data is consistent with that found from this new technique .
the results of this modeling identified brightness enhancements towards the white dwarf and in the threading region . however with this new technique we do not see much enhancement where we expect material to be threaded by the field lines .
the technique of harrop - allin et al .
( 1999a ) was also applied to other observations of hu aqr ( harrop - allin et al .
they found that there was generally enhanced brightness towards the white dwarf and in the threading region , and for some cycles the threading region was significantly brighter .
by comparison to both these methods , vrielmann & schwope ( 2000 ) did not find any enhancement towards the white dwarf , but did identify a brighter threading region .
however , caution should be exercised as they used line emission only , while here we use the total emission ( line and continuum ) .
the results from modeling different light curves of hu aqr may indicate some evolution of the stream brightness with time .
however , the results of applying different models to the same data indicate that the results may be dependent upon the model parameters , as found for the technique of harrop - allin et al .
( bridge et al . 2002 ) . in hu
aqr we see no evidence for emission from where we would expect material to be following a ballistic trajectory . in ep dra however , there is some indication that there is emission from along a ballistic trajectory ( in figures [ fig : fig02 ] and [ fig : fig03 ] the dashed curve represents the ballistic trajectory , seen most clearly in the left - hand panels ) .
the location of material close to the orbital plane in ep dra causes accretion at low latitude and is consistent with the suggested location of the accretion region from the cyclotron models of schwope & mengel ( 1997 ) .
low latitude accretion may be indicative of a more complicated magnetic field geometry , with a quadrupole more likely to cause accretion at lower latitudes ( wu & wickramasinghe 1993 ) .
while the previous techniques all assumed a stream trajectory , here we use a less constrained model that makes few _ a priori _ assumptions about the location of material .
this shows that the trajectory is largely as expected , with the model providing good fits to the light curves of hu aqr and ep dra .
the different results obtained from modeling the same data imply some dependence upon the technique , although this new technique is less sensitive to input parameters than that of harrop - allin et al .
( 1999a ) . a more detailed investigation of the effects of the input parameters and the regularisation curve is needed to fully understand any effects .
we see most of the emission from material close to the white dwarf in both systems , and it is unclear why we see no emission from where we expect material to be following a ballistic trajectory in hu aqr .
the colour information available from s - cam 2 means that we can use the model to determine the colour dependence of the stream emission . in particular
this will help explore the temperature variations along the stream by emphasising which parts of the stream are brightest at bluer wavelengths , and therefore hotter , and where the stream is redder and hence cooler .
we can also apply the model to the consecutive cycles of hu aqr and ep dra , to examine any evolution of the stream structure and temperature .
bridge c.m .
2002 , , 336 , 1129 bridge c.m .
2003 , , accepted charbonneau , p. 1995
, , 101 , 309 hakala p. 1995
, , 296 , 164 hakala p. , cropper m. , ramsay g. 2002 , , 334 , 990 harrop - allin m.k . 1999 , phd thesis , univ .
london harrop - allin m.k .
et al . 1999a , , 302 , 362 harrop - allin m.k .
et al . 1999b , , 308 , 807 harrop - allin m.k . ,
potter s.b . , cropper m. 2001 , , 326 , 788 heerlein c. , horne k. & schwope , a.d .
1999 , , 304 , 145 kohonen t. 1990 , proc . of the ieee , 78 , 1464 kube j. , gnsicke b.t . & beuermann k. 2000 , , 356 , 490 lubow , s.h . &
shu , f.h .
1975 , , 198 , 383 perryman m.a.c .
2001 , mnras , 324 , 899 schwope a.d & mengel s. 1997 , an , 318 , 25 vrielmann s. & schwope a.d .
2001 , , 322 , 269 wu , k & wickramasinghe , d.t .
1993 , , 260 , 141 | we apply a new method of eclipse mapping to the light curves of eclipsing polars .
the technique aims to locate the bright emission associated with the accretion stream , using a technique that makes the fewest prior assumptions about the location of the accretion stream material .
we have obtained data of ep dra and hu aqr with the s - cam 2 superconducting tunnel junction camera using the william herschel telescope .
the location of emission regions in both systems show that previously assumed trajectories are consistent with those found using our technique .
most of the emission is located in a region where we expect material to be confined to magnetic field lines , particularly for hu aqr , while there appears to be less emission from where we conventionally expect material to follow a ballistic trajectory from the @xmath0 point . |
an important goal for our understanding of galaxy formation and evolution is to quantify the state of the molecular gas ( mass , temperature , and density ) available to form stars in young galaxies .
depite many years of study at submm - to - cm wavelengths , very few observational probes of the molecular gas in high - redshift starburst galaxies and active galactic nuclei ( agn ) have emerged .
most efforts have focused on observations of redshifted co line emission , demonstrated to be an effective tracer of the total molecular gas reservoir available to form stars in nearby ultraluminous infrared galaxies ( e.g. downes & solomon 1998 ) .
searches at high - redshift have successfully detected co line emission in more than 60 galaxies ( e.g. solomon & vanden bout 2005 ) , where most have been selected for their luminosity at far - infrared wavelengths , indicating the presence of heavy elements .
high dipole moment molecules such as hcn and hco@xmath8 have been detected in the most luminous objects within this co - selected sample ( solomon et al .
2003 ; vanden bout et al . 2004 ; wagg et al . 2005 ; riechers et al .
these high - density gas tracers have proven very difficult to detect in high - redshift objects , as they are typically an order of magnitude fainter than the corresponding co line emission .
megamaser emission from the oh and h@xmath1o molecules may serve as a powerful alternative means of studying the dense gas in high - redshift objects ( see lo 2005 for a review ) . generally associated with star - formation or agn activity , strong oh or h@xmath1o megamaser emission could be detected at frequencies , @xmath9 10 ghz using current facilities .
the luminosity in the oh maser line has been demonstrated to correlate strongly with fir luminosity ( darling & giovanelli 2002 ) , while a weak correlation may also exist between the fir luminosity and that in the 22 ghz water megamaser line ( castangia et al .
surveys for both molecular species in strongly lensed quasars and submm galaxies ( hereafter smgs ) have been conducted ( wilner et al .
1999 ; ivison 2006 ; castangia et al .
2008 ; edmonds et al . 2009 ) .
the first detection of h@xmath1o megamaser emission at a cosmologically significant distance was in a type - ii qso at @xmath10 by barvainis & antonucci ( 2004 ) .
more recently , impellizzeri et al . ( 2008 ) have detected h@xmath1o megamaser emission in the lensed quasar , mg j0414 + 0534 at @xmath11 .
the discovery of such megamaser emission in a quasar existing when the universe was about 20% of its present age , opens up the possibility that megamaser ( or kilomaser ) emission may also be detectable in equally luminous , star - forming galaxies at these early times .
masers are observed towards star - forming regions in the milky way , and indeed have been detected in a few nearby starburst galaxies ( e.g. haschick & baan 1985 ; henkel , wouterloot & bally 1986 ; hagiwara , diamond & miyoshi 2002 , 2003 ; nakai , sato & yamauchi 2002 ) .
extragalactic h@xmath1o megamasers with isotropic luminosities , @xmath1210 l@xmath3 , are confined to the nuclei of galaxies , where the physical conditions needed to reach population inversion , namely temperatures above 300 k and densities greater than 10@xmath13 @xmath14 , are more likely to be present . among the most luminous class of high - redshift starburst galaxies
are the submm galaxies ( smgs ) , discovered in blank - field submm / mm - wavelength bolometer surveys with single - dish telescopes ( smail , ivison & blain 1997 ; hughes et al .
1998 ; barger et al.1998 ; bertoldi et al .
these typically have 850 @xmath15 m flux densities , @xmath16 mjy , implying fir luminosities , @xmath1710@xmath18 @xmath19 ( assuming a greybody distribution with dust temperature , @xmath20 k and @xmath21 describes the fir - to - mm emission ) .
optical spectroscopic surveys have published redshifts for less than 100 smgs ( e.g. chapman et al .
2003 , 2005 ) .
these surveys have been hindered by the coarse angular resolution of submm / mm telescopes , so that deep radio interferometry has been needed to identify the correct multi - wavelength counterparts , and also by the faintness of the optical / infrared counterparts ( e.g. wang et al .
an alternative method has been to use broad bandwidth mm - to - cm wavelength receivers in searching for redshifted co line emission ( wagg et al . 2007 ;
daddi et al . 2009a , b ) .
these searches have so far been moderately successful in obtaining redshifts for @xmath22 smgs .
lower frequency searches for redshifted megamaser and gigamaser emission may prove to be a viable alternative to searches for molecular co line emission in smgs ( townsend et al . 2001 ; ivison et al .
2006 ; edmonds et al . 2009 ) , with telescopes like the expanded very large array ( evla ) .
such observations could also provide an effective means of probing the dense gas in these objects , since previous searches for hcn emission in smgs have not been successful ( carilli et al .
2005 ; gao et al .
2007 ) . in this study
, we present evla and arecibo observations of h@xmath1o megamaser emission in two gravitationally lensed smgs . throughout this work ,
we assume h@xmath23=71kms@xmath6 , @xmath24@xmath25=0.27 , and @xmath24@xmath26=0.73 ( spergel et al . 2007 ) .
the two smgs in our sample were discovered in submm surveys of low redshift galaxy clusters , where gravitational lensing has been used to constrain the faint end of the submm / mm source counts ( @xmath272 mjy ; e.g. smail , ivison & blain 1997 ; knudsen et al .
both sources have also been previously detected in co line emission , so that the emission redshift of the molecular gas region is known with sufficient accuracy to conduct further searches for redshifted h@xmath1o megamaser emission within the narrow spectral bandwidth afforded by the current vla correlator .
previous studies have observed h@xmath1o megamaser emission in high - redshift agn , whereas our sample is comprised of purely starburst galaxies . at @xmath28 ( barger et al .
1999 ) was discovered in the 850 @xmath15 m scuba survey of the a1835 cluster field ( ivison et al .
2000 ) , and subsequently detected in line emission by frayer et al .
the lensing amplification factor of this source has been the subject of some debate , and may be as high as @xmath2925 ( downes & solomon 2003 ; motohara 2005 ; smail et al . 2005 ; smith et al .
2005 ) , however in this analysis we adopt a lower factor of 3.5 ( smail et al .
2005 ) so as to be conservative in our derived luminosity limits in the case of a non - detection .
the optical spectral characteristics of this smg are consistent with that of a starburst galaxy ( frayer et al .
1999 ) , while the mid - ir spectrum is also consistent with no agn being present in this smg ( rigby et al .
our second target , is a @xmath30 smg lensed into three components by the a2218 cluster ( kneib et al .
2004 ) , with the brightest component lensed by a factor of @xmath2922 .
all three components have been detected in co line emission ( sheth et al .
2004 ; kneib et al .
2005 ) , revealing a mass in molecular gas of only @xmath293@xmath3110@xmath32 m@xmath3 .
non - detections of in both the 0.52 and 28 kev _ chandra _ x - ray bands imply that no obscured agn is present , a conclusion supported by the mid - ir spectrum of smm j16359 + 6612 , which is similar to a starburst galaxy template derived from @xmath33 galaxies ( rigby et al .
the rest frequency of the h@xmath1o maser line is 22.23508 ghz , so that at the redshifts of our two targets ( @xmath0 ; table 1 ) , the extended frequency coverage provided by the new evla c - band receivers ( 48 ghz ) is required to conduct our search . observations of were obtained in two five - hour tracks on september 28 and 29 , 2008 ( project aw750 ) .
a total of 16 evla antennas were available during a move from the compact d to the most extended a configuration .
another five - hour track in the b configuration was observed toward on february 17 , 2009 ( project aw761 ) , when 18 evla antennas were available . in both cases ,
the central tuning frequency was dictated by the redshift of the co lines detected previously in these two sources ( frayer et al .
1999 ; sheth et al .
a bandwidth of 6.25 mhz was used to cover the velocity range of the co lines , and a spectral resolution of 97.656 khz was adopted .
flux density and bandpass calibration were performed using observations of 3c286 , while we observed 1642 + 689 as a phase calibrator for smm j16359 + 6612 , and 1354 - 021 for smm j14011 + 0252 .
the data were reduced using the nrao s astronomical image processing system ( aips ) .
table 1 summarizes our evla observation parameters .
we have also observed j14011 + 0252 with the 305 m arecibo radio telescope , in order to search for h@xmath1o megamaser emission at a frequency of 6.237 ghz .
these data were obtained during 3.5 hours of observing on march 12 and 13 , 2009 , with the c - high receivers and using the standard ` on / off ' observing mode .
both orthogonal polarizations were recorded .
five minutes were spent in each of the ` on ' and ` off ' positions during a single scan .
the correlator was configured to provide 64 channels of spectral resolution , and the final rms per channel is 152 @xmath15jy / beam after hanning smoothing .
these observations are @xmath2920% more sensitive than our evla observations of the same source .
the data were analyzed using standard arecibo observatory _
_ routines .
our c - band spectra of and do not reveal any evidence for h@xmath1o line emission . given the strong lensing amplification of these objects by foreground galaxy clusters , our spectra can be used to derive very deep limits to the line luminosities of any h@xmath1o megamasers which may be present . following edmonds et al .
( 2009 ) , we first correct the spectrum extracted at each of the three components of for its corresponding lensing factor ( kneib et al . 2005 ) , before adding these together with the appropriate weighting to create the average spectrum shown in figure [ fig : fig1 ] .
the rms of this average spectrum is 11.6@xmath15jy per 97.7 khz channel , which is a factor of @xmath34 lower than that obtained by edmonds et al .
we decide to use the more sensitive arecibo spectrum in calculating our h@xmath1o line luminosity limits , which is plotted in figure [ fig : fig1 ] .
the lensing - corrected rms of this arecibo spectrum is 43.4@xmath15jy per 97.7 khz .
our line luminosity limits are calculated for a range in h@xmath1o megamaser linewidths , 20 , 40 , 60 , and 80 km s@xmath6 , consistent with that of nearby fir luminous galaxies ( henkel et al .
we then calculate lensing - corrected 3-@xmath4 limits to the isotropic line luminosity in smm j16359 + 6612 , @xmath35 946 , 1339 , 1639 , and 1893 @xmath19 .
similarly , in the isotropic line luminosity limits are @xmath35 3735 , 5282 , 6469 , and 7470 @xmath19 for our assumed linewidths of 20 , 40 , 60 and 80 km s@xmath6 .
a previous search for h@xmath1o megamaser emission in has been conducted by edmonds et al .
( 2009 ) . while the sensitivity achieved here is nearly a factor of three higher , we do not find any evidence for line emission . assuming a dust temperature , @xmath36 k , the corrected 850 @xmath15 m flux density of is consistent with a fir luminosity , @xmath37 l@xmath3 , which can be used to predict the expected h@xmath1o megamaser line luminosity according to the relationship observed in star - forming regions ( genzel & downes 1979 ; jaffe et al .
the predicted line luminosity is , @xmath38 l@xmath3 , consistent with our derived limits . in the case of smm
j14011 + 0252 , the corrected 850 @xmath15 m flux density would imply , @xmath39 l@xmath3 , for the same assumptions on the dust temperature .
we should then expect an h@xmath1o megamaser luminosity , @xmath40 l@xmath3 , which is also consistent with our observed limits . if we assume the weak correlation between the h@xmath1o megamaser luminosity and fir luminosity measured for fir bright galaxies by castangia et al .
( 2008 ) , we would expect megamaser line luminosities of , 3800 and 16800 l@xmath3 for and smm j14011 + 0252 , respectively .
the sensitivity of our constraints on h@xmath1o megamaser emission in these two star - forming ultraluminous infrared galaxies ( ulirgs ) would suggest that they do not contain strong megamasers , such as that observed in mg j0414 + 0534 at @xmath41 .
although our sample size is small , these non - detections in strongly lensed , metal and gas - rich starburst galaxies would imply that strong h@xmath1o megamaser emission may not be common among such objects . as pointed out by edmonds et al .
( 2009 ) , it is likely that the physical conditions of the interstellar gas in these predominantly star - forming smgs is not sufficiently warm and dense to excite megamaser line emission , which is more readily excited in the x - ray irradiated environments close to a central agn ( neufeld et al . 1994 ) .
another possibility is that the density and temperature of the gas are large , but the gas is not sufficiently coherent in velocity with the maser source to produce strong amplification .
weiss et al .
( 2005 ) observe multiple transitions of co line emission in smm j16359 + 6612 , which reveal a double - peaked co line profile , consistent with either a circumnuclear toroid of rotating molecular gas , or molecular gas associated with separate components of an ongoing merger .
their excitation analysis finds the gas to be cool , with temperatures most likely below @xmath2980 k , while the large star - formation rate ( @xmath29500 m@xmath3 yr@xmath6 ) inferred by kneib et al .
( 2004 ) should result in a high gas excitation if the co line emission were due to a single , massive component .
this low excitation gas spread over multiple components is also consistent with our non - detection of an h@xmath1o megamaser in smm j16359 + 6612 . although our luminosity limits are consistent with what one would expect from a simple extrapolation of the @xmath42 relation in galactic star - forming regions , these limits do not rule out the presence of h@xmath1o `` kilomasers '' ( @xmath43 l@xmath3 ) , which are typically found both within , and exterior to the nuclei of nearby star - forming galaxies ( e.g. hagiwara et al . 2001 , 2003 ; henkel et al .
2004 ) . detecting such emission in a member of the , @xmath445 mjy , smg population would require hundreds of hours of observing time using the full evla , or arecibo .
nearby ulitraluminous infrared galaxies , which are the nearest analogues to the high - redshift smgs , generally do not exhibit strong water megamaser emission .
the exceptions to this are ngc6240 and ugc5101 ( hagiwara et al .
2002 ; zhang et al .
2006 ) , which are both believed to contain agn .
such an association of strong h@xmath1o megamaser emission with nearby agn , rather than starburst ulirgs , is consistent with our non - detections in and smm j14011 + 0252 .
this would also support the original claim by edmonds et al .
( 2009 ) that h@xmath1o megamasers will not be an effective means of measuring redshifts for starburst smgs in future evla surveys .
although we do not detect megamaser emission in the two starburst galaxies observed , the detection of an h@xmath1o megamaser in mg j0414 + 0534 suggests that such emission may also be present in smgs hosting an agn .
the results of deep x - ray imaging of smgs shows that 2850% may harbour an agn ( alexander et al .
2005 ) , while the median mid - infrared spectrum of a sample of 24 smgs ( menndez - delmestre et al .
2009 ) , suggests that less than one third of the fir luminosity is likely powered by such agn ( pope et al .
we can use these results to make simple predictions for future c - band surveys with the evla . with the already existing , expanded frequency coverage of the evla c - band receivers , it is possible to search for water megamaser emission in @xmath45 smgs .
the forthcoming availability of the widar correlator will also make it possible to obtain spectroscopy over the entire 48 ghz window , so that searches for redshifted megamaser emission could be conducted over the redshift range , @xmath46 , within a field of view of 44 sq .
arcmins in size .
the best current constraints on the 850 @xmath15 m number counts predict about 0.1 smgs with @xmath476 mjy , per sq .
arcmin ( coppin et al . 2006 ) , so that we would expect 4.4 bright smgs in a single c - band pointing .
assuming @xmath20 k and @xmath48 for this sub - sample of the population , then their expected fir luminosities would be , @xmath49 l@xmath3 .
roughly 73% of the bright , radio - detected smg population ( chapman et al . 2005 ) , would have a redshift that places the 22.23508 ghz water maser line in the frequency interval covered by the evla c - band . while it may be that the 15 - 20% of bright smgs with no 1.4 ghz radio counterpart are at redshifts , @xmath50 ( e.g. ivison et al . 2005 ; wagg et al . 2009 ) , we will assume here that all of these are at redshifts , @xmath51 ( although see coppin et al . 2009 ) .
if we make the assumption that 30% of the smgs in this redshft range have agn , and that 30% of their fir luminosity is powered by this agn , then we would expect one smg in a single c - band pointing to have agn with fir luminosities , @xmath52 l@xmath3 . if we assume the ratio between @xmath53 and l@xmath54 measured in mg j0414 + 0534 ( impellizzeri et al .
2008 ) , and also that these smgs do indeed host a water megamaser where the accretion disk is being viewed edge - on , then they could exhibit isotropic h@xmath1o line luminosities , @xmath55 l@xmath3 . for the faintest ,
unlensed sources at @xmath56 , a maximum of 30 evla hours would be needed to detect the megamaser emission .
however , we note that no unlensed megamaser has been detected with such an extreme luminosity , and it is possible that a cutoff in the luminosity function of h@xmath1o megamasers exists at high luminosities . based on their detection in mg j0414 + 0534 , impellizzeri et al .
( 2008 ) argue that the probability of finding an h@xmath1o megamaser in a single , pointed observation of a high - redshift object is 10@xmath57 for a non - evolving line luminosity function , and 0.05 if this function evolves strongly with redshift , like ( 1 + @xmath58)@xmath59 . therefore ,
`` blind '' evla surveys for redshifted megamaser emission will serve to constrain any evolution in the luminosity function .
we have used the evla and arecibo to search for h@xmath1o megamaser emission in two strongly lensed , starburst smgs at @xmath0 .
the gravitational lensing of these two objects allows us to rule out the presence of strong megamaser emission , such as that observed in mg j0414 + 0534 at @xmath41 .
we are not able to rule out the presence of kilomaser emission commonly associated with star - forming galaxies .
these are the deepest limits to date on the luminosities of h@xmath1o megamasers in high - redshift galaxies .
it is possible that future c - band surveys with the evla will detect one h@xmath1o megamaser emitting smg within a single , evla pointing .
currently planned evla observations of smgs with observed agn characteristics , and existing co emission line data , will allow us to better assess the success rate of conducting such `` blind '' megamaser surveys in the future . at the time of completion
, the evla will have been outfitted with new x - band ( 8 - 12 ghz ) and ku - band receivers ( 12 - 18 ghz ) , so that it will also be possible to conduct searches for h@xmath1o megamasers at lower redshifts .
we thank the nrao staff involved in the evla project for making these observations possible .
jw is grateful for support from the max - planck society and the alexander von humboldt foundation .
we would also like to thank david wilner , chris carilli , karl menten , liz humphries , david hughes and robert edmonds for helpful discussions .
lcc @xmath60 : & 2.517 & 2.565 + frequency : & 6.3221 ghz & 6.2365 ghz + pointing center ( j2000 ) ra : & 16@xmath6135@[email protected] & 14@xmath6101@[email protected] + dec : & + 66@xmath6412@[email protected] & + 02@xmath6452@[email protected] + synthesized beam : & @xmath65 , p.a .
36.4@xmath66 & @xmath67 , p.a .
@xmath68 + channel spacing : & 97.7 khz & 97.7 khz + line rms ( 97.7 khz channel ) : & 221 @xmath15jy beam@xmath6 & 275 @xmath15jy beam@xmath6 + | we present expanded very large array and arecibo observations of two lensed submm galaxies at @xmath0 , in order to search for redshifted 22.235 ghz water megamaser emission .
both smm j14011 + 0252 and smm j16359 + 6612 have multi - wavelength characteristics consistent with ongoing starburst activity , as well as co line emission indicating the presence of warm molecular gas .
our observations do not reveal any evidence for h@xmath1o megamaser emission in either target , while the lensing allows us to obtain deep limits to the h@xmath1o line luminosities , @xmath2 l@xmath3 ( 3-@xmath4 ) in the case of smm j14011 + 0252 , and @xmath5 l@xmath3 for smm j16359 + 6612 , assuming linewidths of 80 km s@xmath6 .
our search for , and subsequent non - detection of h@xmath1o megamaser emission in two strongly lensed starburst galaxies , rich in gas and dust , suggests that such megamaser emission is not likely to be common within the unlensed population of high - redshift starburst galaxies .
we use the recent detection of strong h@xmath1o megamaser emission in the lensed quasar , mg j0414 + 0534 at @xmath7 to make predictions for future evla c - band surveys of h@xmath1o megamaser emission in submm galaxies hosting agn . |
dark energy and dark matter appear as two of the most important questions of modern cosmology .
many models trying to answer these questions involve scalar fields . in particular , to explain the behavior of dark energy , quintessence models @xcite consider scalar fields with potentials dominating today , giving birth to a negative pressure .
but dark matter can also be explained through scalar fields , as in boson star models @xcite or scalar field dark matter models @xcite .
a unification of dark matter and dark energy can also be performed into a dark fluid @xcite , which can again be modeled with a dark fluid scalar field @xcite .
thus , scalar fields are involved in many cosmological models .
+ one of the main difficulties to build scalar field - based models is the choice of a potential .
the usual way to deal with scalar fields in cosmology is to consider them as classical , and to study their evolution .
however , it is known that quantum fluctuations could modify the potential .
+ in this paper , we study the alterations of real or complex scalar field potentials due to quantum corrections .
we focus on the one - loop quantum corrections to the original potentials , while computing effective potentials . we will therefore be able to test the validity of several dark energy , dark matter and dark fluid potentials .
a similar study has been performed on several quintessence real scalar field models @xcite , and we extend here that study to a broader range of cosmological models containing complex scalar fields or two coupled real scalar fields .
+ + the cosmological action is written as : @xmath0 where @xmath1 is the associated lagrangian density .
we consider that the lagrangian contains a curvature term leading to einstein equations , a term involving scalar fields minimally coupled to gravity , and a term containing fermions potentially coupled to the scalar fields @xcite .
we have then @xmath2 the curvature lagrangian is @xmath3 where @xmath4 is the planck mass and @xmath5 the curvature tensor . for a model containing a complex scalar field @xmath6 , the scalar field lagrangian can be written as : @xmath7 where @xmath8 is the potential of the field . for two real scalar fields @xmath9 and @xmath10 ,
the scalar field lagrangian writes : @xmath11 -v(\phi_1(x),\phi_2(x ) ) \;\;.\ ] ] these two cases are completely equivalent , and the complex scalar field can be written in function of the two real scalar fields : @xmath12 in presence of a single fermionic species coupled to the scalar fields , the fermionic lagrangian reads : @xmath13 \psi(x)\;\;,\ ] ] where @xmath14 is the fermion mass , which is dependent on @xmath6 , @xmath15 s are dirac matrices and @xmath16 is the covariant derivative .
+ we first restrict ourselves to the flat minkowski space , so that @xmath17 , @xmath18 , @xmath19 and @xmath20 . + the method we use here to study the quantum corrections is the saddle point expansion described in @xcite .
the effective action of the scalar field writes , to higher order : @xmath21= \int d^4x \ , \mathcal{l}_1[\phi_{cl } ] + \frac{i}{2 } \log \det\left[\frac{\delta^2 \mathcal{l}_1}{\delta \phi_i \ , \delta\phi_j}\right ] + \cdots \;\;,\ ] ] where @xmath22 is the renormalized part of @xmath23 , the subscript @xmath24 indicates that the fields are classical , and @xmath25 .
the effective potential can be written in function of @xmath26 $ ] such that @xmath27 \;\;,\ ] ] where @xmath28 is a space - time volume .
we consider here only the one - loop corrections to the potential as shown in fig [ fig ] . and
@xmath29 in the potential.,title="fig:",height=151 ] + + and @xmath29 in the potential.,title="fig:",height=75 ] + ( a)(b ) we need to define the effective masses @xmath30 and
@xmath31 with a large momentum cutoff @xmath32 , and ignoring the @xmath6-independent contributions and the graphs of higher orders , the calculation of the effective potential due to the leading order scalar loop for two real coupled scalar fields leads to : @xmath33 + \frac{m_b^4 } { 32\pi^2 } \left[\ln\left(\frac{m_b^2}{\lambda}\right)-\frac{3}{2}\right]\;\;.\ ] ] if the terms proportional to @xmath34 and @xmath35 are important in the context of usual field theory , in the case of cosmological scalar fields the potential is of the order of @xmath36 @xmath37 , and therefore we can safely disregard these terms in comparison to the terms proportional to @xmath38 and @xmath39 . for this reason , even if many of the cosmological potentials are non - renormalizable in the strict sense of field theory , we can consider that the cosmological potentials are renormalizable as the higher order terms are so small .
therefore , the effective potential reduces to : @xmath40 we can also calculate the correction due to fermions : @xmath41 ^ 2 - \frac{[m_f(\phi_{cl})]^4 } { 32\pi^2 } \left[\ln\left(\frac{[m_f(\phi_{cl})]^2}{\lambda_{ferm}}\right)-\frac{3}{2}\right ] \;\;.\ ] ] again , because the fermion mass is very small in comparison to the planck mass , the terms proportional to @xmath42 ^ 4 $ ] can be neglected .
we obtain finally the one - loop effective potential for two real coupled scalar fields : @xmath43 ^ 2 \;\;,\ ] ] or , for a single complex scalar field : @xmath44 ^ 2 \;\;.\ ] ] these results are in agreement with the single real scalar field results of @xcite . + we disregarded here the higher order terms , which is a reasonable approximation , as the potential and its derivatives are today of the order of the critical density , _
i.e. _ of the order of @xmath45
. we also note that the ignored @xmath6-independent contributions would lead to a cosmological constant of the order @xmath46 , which is by far much larger than the critical density .
this well - known problem appearing in the majority of field theories could hopefully be solved by some mechanism which could make these contributions vanish .
+ in the following , we will consider that @xmath47 unless stipulated otherwise , and we use units in which @xmath48 .
we study first the case in which the scalar fields are not coupled to fermions , for quintessence , scalar field dark matter and scalar field dark fluid models , and then we consider the case with a coupling to the fermions .
we calculate in this section the one - loop corrections to several cosmological scalar field potentials .
first , we consider a single real scalar field associated to the sugra potential as in a dark energy quintessence model @xcite .
then , we continue with the case of two coupled real scalar fields , involved in the coupled quintessence model @xcite .
we also discuss a dark matter model based on a complex scalar field associated to a quadratic @xcite or quartic @xcite potential .
finally , we analyse the case of a unifying dark fluid model involving a complex scalar field associated to a potential containing a quadratic term and a decreasing exponential term @xcite .
the sugra potential for a quintessence real scalar field reads @xcite : @xmath49 where @xmath50 .
we obtain the effective one - loop potential : @xmath51 \;\;.\ ] ] it can be noted that the shape of the potential is changed .
however , the global @xmath52 factor can be reabsorbed into the @xmath53 factor : @xmath54 , and the @xmath55 factor can also disappear with the redefinition @xmath56 .
thus , the shape can remain similar if @xmath57 for a usual @xmath58 , as the present value of the field is of the order of @xmath4 , this inequality is respected today .
however , as already noticed in @xcite concerning inverse power law potentials , problems could arise earlier as the correction can strongly modify the form of the potential .
fortunately , the argument developed in @xcite should still hold for this kind of potential : the corrections are important only when quintessence was subdominant , and as corrections contains only negative power of @xmath59 , the field should be able to roll down the potential , and when the field becomes relevant to account for dark energy , the corrections may have become negligible . a more detailed analysis is nevertheless needed to check the validity of this scenario .
we consider now the potential for two coupled real scalar fields of @xcite : @xmath60 with @xmath61 the derivatives of the potential reads : @xmath62 \exp(-\lambda \phi_1 ) \;\;,\ ] ] and @xmath63 so that the effective potential becomes : @xmath64 using the same values for the parameters as in @xcite , we can show that the additional terms which can not be reabsorbed are presently of the same order as the value of the non - corrected potential .
therefore , the structure of the polynomial is globally changed , and unfortunately , the additional terms can not be reabsorbed in the constants .
a way out could be to replace @xmath65 by a more general polynomial of the form @xmath66 the @xmath67 being constants . in this case
, the loop correction terms can be reabsorbed into the constants .
such a possibility has of course to be studied further , but would probably lack of credibility due to the high number of independent parameters .
we study also the following potential of a complex scalar field @xcite : @xmath68 the effective potential can be obtained : @xmath69 as the effective potential is defined up to a constant , the constant term is irrelevant and can be safely disregarded . for @xmath70 ,
the potential is therefore unchanged . for @xmath71 ,
the potential is still stable , because the term @xmath72 can be reabsorbed into the redefinition @xmath73 .
thus , this potential is stable through loop corrections , and is in fact renormalizable .
of course , these results were predictable , as the quadratic and quartic potentials are well - known potentials which have been intensively studied . finally , we consider the potential @xcite : @xmath74 this potential can lead to both dark matter and dark energy behaviors .
the one - loop effective potential reads : @xmath75 again , the constant term can be safely disregarded .
the term @xmath76 can be reabsorbed in the redefinition @xmath77 .
however , a problem arises from the extra term @xmath78 which modifies the potential . considering the values of the parameters derived in @xcite
, this term can not be neglected today and would lead to a strong modification of the potential , unless @xmath79 .
we can however note that at earlier epochs , as @xmath80 was larger , this term was extincted by the exponential .
+ + a way to solve this problem would be to modify the potential such as : @xmath81 in this case , the effective potential writes : @xmath82 \exp(-\beta \phi_{cl}^\dagger\phi_{cl})\nonumber\;\;.\end{aligned}\ ] ] the constant term is irrelevant . with two redefinitions :
@xmath83 and @xmath84 , the correction terms of the second line vanish , but the last correction term of the first line still remains and modifies the potential shape .
however , for @xmath85 , this term is negligible in comparison to the other terms , and the shape of the potential is then conserved .
if @xmath86 and @xmath87 is small in comparison to @xmath88 , the analysis of @xcite would remain valid . in the case of a larger @xmath87
, we could modify again the potential , such as : @xmath89 where @xmath90 goes from 0 to @xmath91 . in this case
, the corrections can be reabsorbed into the constants @xmath92 , and the last correction of order @xmath93 will hopefully be negligible today in comparison to the other terms .
yet , such a potential would lead to a behavior similar to that described in @xcite , provided the redefined terms @xmath94 ( for @xmath95 ) are today small in comparison to @xmath96 .
+ + to conclude this section , we have studied the one - loop corrections to several scalar field potentials .
some potentials can have difficulties to resist the quantum fluctuations .
however , it is possible to build more suitable and robust potentials , in particular for the dark fluid model .
we will now consider the corrections due to a coupling of the scalar fields to fermions .
the fermion term in the effective potential reads : @xmath97 ^ 2 \;\;.\ ] ] if the fermion mass is @xmath6-independent and of the order of 100 gev , and considering that the fermionic cutoff is taken at gut scale : @xmath98 , the fermionic term is of the order of @xmath99 , which is extremely large in comparison to the present value of the potential @xmath100 .
thus , as noticed in @xcite , the fermionic corrections do not dominate the potential only if @xmath42 ^ 2 $ ] takes a form which is already contained in the classical potential , and which can then be reabsorbed in the constants of the potential .
+ + let us focus for example on the dark fluid potential of eq .
( [ df2 ] ) .
the fermion mass can then take a form such as : @xmath101 or the square root of a sum of similar terms .
however , a simpler acceptable form would be : @xmath102 ^ 2=(m_f^0)^2 \phi_{cl}^\dagger \phi_{cl } \;\;.\ ] ] in this case , @xmath103 can be reabsorbed by the redefinition @xmath104 .
+ + we now consider the case where the coupling does not share the shape of the potential , and where the one - loop corrections can not be reabsorbed by higher order corrections .
we can write the fermion mass as @xmath105 assuming here that @xmath106 , we have : @xmath107 \;\;.\ ] ] the constant term is pointless , as the effective potential is defined up to a constant .
thus , the global potential has a similar structure if @xmath108 for a present value of the potential such as @xmath100 , this leads to a limit on the present value of @xmath109 : @xmath110 this limit is so stringent that it nearly forbids a dependence of the mass on @xmath111 if the shape of the coupling is different from the shape of the potential .
this result is similar for the quintessence potentials , and is in agreement with @xcite .
we however have to notice that as we restricted ourselves to one - loop corrections , this limit has to be interpreted with reserve , as the higher order corrections could also modify the structure of the effective potential .
we have just considered the effective one - loop quantum corrections to the scalar field potential .
however , other corrections can also have an important influence on the scalar field behavior . in this section ,
we study the corrections due to the space - time curvature , and the corrections to the kinetic term . to simplify , we discuss here the case of a single real scalar field . to study the effects of the curvature on the potential
, @xmath112 is assumed .
we consider the same lagrangian densities as in the introduction .
using the asympotic expansion of the heat kernel for the involved operators , one can show that the corrections involving the curvature tensor which modify the dynamics of the scalar field read @xcite : @xmath113\\ \nonumber & & - \frac{1}{32\pi^2 } \left ( [ m_f(\phi_{cl})]^2 + \frac{r}{12 } \right)^2\left[\ln\left(\frac{[m_f(\phi_{cl})]^2+r/12}{\lambda_{ferm}}\right)-\frac{3}{2}\right ] \;\;,\end{aligned}\ ] ] where @xmath114 we can first note that in the case of a minkowskian space @xmath115 , the logarithmic terms of equations ( [ vsca_tot ] ) and ( [ vfer_tot ] ) are retrieved . practically , the other correction terms are not modified by the curvature , and only the logarithmic terms that we disregarded receive contributions . therefore , for small curvature , _
i.e. _ @xmath5 of the order of @xmath116 or @xmath117 , the terms which are involved here have a negligible impact on the effective potential at cosmological scales , and can be again neglected .
the conclusion can be generalized to the case of two real or one complex scalar fields . however , more attention is of course needed in the case of a larger curvature .
the analyzes of the previous sections are valid provided the khler potential is flat . at low energy , under a scale @xmath118 , the khler potential may be a more complicated function , called @xmath119 , which can be expanded in taylor series .
for a single real scalar field , it writes : @xmath120 where @xmath121 are constants . if @xmath122 , the khler potential is considered as flat , and therefore this expansion holds if @xmath123 .
+ in a flat space - time , the scalar field low - energy lagrangian is @xmath124 where @xmath125 is the khler metric on the one dimensional curve defined as : @xmath126 to better render the physical meaning of the khler potential , we redefine the field such as @xmath127 integrating and using an iterative inversion method , we can show that @xmath128 where the @xmath129 are constants which can be determined iteratively . with this redefinition , the kinetic term of @xmath6 is transformed into a standard kinetic term for @xmath130 , and the potential writes : @xmath131 where @xmath132 denotes the @xmath133-th derivative of the potential @xmath8 , and @xmath134 are calculable constants .
+ thus , the additional terms to the effective potential due to the kinetic term corrections are : @xmath135 we can first recall that the corrections appear if @xmath123 . therefore , the energy scale @xmath118 has an important influence here .
for our study , we can assume that @xmath118 is of the order of the planck mass .
the correction terms appear as powers of the scalar field times derivative of the potential . for the potentials we studied , because they involve either polynomials or exponential terms , they will keep their initial structure , but receive corrections under the form of powers of the scalar fields . + for the dark matter and dark fluid potentials ,
as the kinetic corrections are small , they should not perturb much the dynamics of the scalar fields , at most providing some small long - range effects .
however , in the case of quintessence models , a more detailed study would be needed to determine if these corrections could perturb the tracking properties , and then the behavior the fields .
+ to conclude this sections , we have seen that other corrections can perturb the scalar fields .
however , in models with small curvature or nearly flat khler potential , the considered corrections can be neglected .
in this letter , we performed the calculation of the one - loop effective potentials of several cosmological models based on scalar fields .
we have seen that some potentials are stable against the quantum fluctuations , and in particular the simple quadratic and quartic potentials of the dark matter scalar field models .
different solutions to the instabilities of the coupled scalar field quintessence potential and of the dark fluid potential have been proposed .
+ focusing on the fermionic corrections , we have confirmed the results of @xcite with regard to the extreme restriction of the coupling to the fermions .
this constraint is relaxed if the coupling has a common structure with the scalar field potential or can be compensated by higher order corrections .
+ concerning corrections due to the curvature or the kinetic term , we have seen that in usual case they are small , but they may have to be carefully studied in the case of strong curvature of the space - time or the khler potential .
+ the calculations we performed in this work concern only the one - loop order corrections , and therefore do not give access to the true effective potential .
however , such studies reveal to be valuable tools to test and build robust scalar field potentials .
a.a.s research is partially funded by a bqr grant from universit lyon 1 , and he acknowledges the support from the horizon project as well . f.m . acknowledges the support from nserc .
kiselev , class .
* 22 * ( 2005 ) , 541 ; + f.s .
guzman & l.a .
urena - lopez , phys .
d * 68 * ( 2003 ) , 024023 ; + a. bernal & f.s .
guzman , phys .
d * 74 * ( 2006 ) , 063504 ; + a. bernal & f.s .
guzman , phys .
d * 74 * ( 2006 ) , 103002 . | we study the loop corrections to potentials of complex or coupled real scalar fields used in cosmology to account for dark energy , dark matter or dark fluid .
we show that the sugra quintessence and dark matter scalar field potentials are stable against the quantum fluctuations , and we propose solutions to the instability of the potentials of coupled quintessence and dark fluid scalar fields .
we also find that a coupling to fermions is very restricted , unless this coupling has a structure which already exists in the scalar field potential or which can be compensated by higher order corrections . finally , we study the influence of the curvature and kinetic term corrections .
+ + pacs numbers : 98.80.-k , 11.10.-z , 95.35.+d , 95.36.+x mta - phys-0701 * one - loop quantum corrections to cosmological scalar field potentials * a. arbey + _ universit de lyon , lyon , f-69000 , france ; universit lyon 1 , villeurbanne , f-69622 , france ; centre de recherche astrophysique de lyon , observatoire de lyon , 9 avenue charles andr , saint - genis laval cedex , f-69561 , france ; cnrs , umr 5574 ; + ecole normale suprieure de lyon , lyon , france _ + f. mahmoudi + _ department of physics , mount allison university , 67 york street , sackville , + new brunswick , canada e4l 1e6 _ |
the classical occupancy problem is one of the cornerstones of discrete probability , dating back to its early ages ( and hence encountered over and over again by the generations of students studying elementary probability through the evergreen hits like the birthday problem , the coupon collector s problem , etc.@xcite ) .
it still attracts lots of research interest , especially in recent years , mainly due to its numerous applications spreading across the board , from sampling statistics and quality control to quantum physics , bioinformatics and computer science . for an introduction to the field and a survey of the many models and results ,
see @xcite and further references to original work therein . in this paper
, we are concerned with a version of the occupancy problem in an _ infinite urn scheme _ ( first considered by bahadur @xcite and later on studied by darling @xcite and most systematically by karlin @xcite ) , in which the balls labeled @xmath2 are thrown independently at an infinite array of boxes ( _ urns _ ) @xmath3 , with fixed probability ( _ frequency _ ) @xmath4 of hitting box @xmath5 .
the frequencies @xmath4 are assumed to be strictly positive and satisfying @xmath6 without loss of generality , we further assume that the sequence @xmath7 is non - increasing , @xmath8 .
let @xmath9 be the number of boxes discovered by the first @xmath0 balls ( i.e. , occupied by at least one of the first @xmath0 balls ) .
many other interpretations of this functional appear in the literature : for instance , when @xmath7 is considered as a probability distribution on positive integers , @xmath9 is the number of distinct values occurring among @xmath0 random values sampled independently from @xmath7 .
since there are infinitely many boxes , @xmath9 increases unboundedly ( with probability one ) as more balls are thrown , which also implies ( e.g. , by fatou s lemma ) that the same is true for the expected number of occupied boxes , @xmath10 .
moreover , as shown by karlin ( * ? ? ?
* theorem 8) , @xmath11 with probability one ( an earlier result about convergence in probability was obtained by bahadur @xcite ) .
the more delicate asymptotic properties of the random variable @xmath9 are largely determined by its variance @xmath12 .
it is known @xcite that the distribution of @xmath9 converges to a normal distribution provided that @xmath13 as @xmath14 .
the latter occurs , for instance , when the frequencies have a power - like decay , @xmath15 ( @xmath16 ) with @xmath17 or , more generally , satisfy a condition of regular variation @xcite .
( here and throughout , @xmath18 stands for a generic positive constant , specific value of which is not important . ) in this paper , we essentially focus on the opposite situation , that is , when @xmath1 is uniformly bounded ( and hence the distribution of @xmath9 does not converge to normal ) . in particular , we prove the following surprising characterization of frequencies @xmath7 for which the variance @xmath1 tends to a finite limit as @xmath14 .
[ thm : limit ] a finite limit @xmath19 exists if and only if for some integer @xmath20 the frequencies satisfy the `` lagged ratio '' condition @xmath21 and in this case the limiting value @xmath22 coincides with the lag @xmath23 .
the striking consequence of this result is that whenever the finite limit of the sequence @xmath24 exists , it must be a positive integer number , @xmath25 .
the issue of converging variance was first queried in the seminal paper by karlin @xcite , where in particular he appreciated as `` formidable if not impossible '' the task to determine the behavior of the variance @xmath1 without some regularity assumptions . in particular , adopting the condition of regular variation of the frequency tail , he came up with a sufficient condition for the existence of a finite limit of @xmath1 ( * ? ? ?
* theorem 2 ) .
in fact , as we shall see below ( in section [ sec5 ] ) , convergence to a finite limit , combined with the special dyadic structure of the counting measure controlling the frequency input , is a regularity condition in itself , being strong enough to ensure the result of theorem [ thm : limit ] .
( to be more precise , the `` dyadic '' feature mentioned above , pertains primarily to the _ poissonized _ version of the problem , i.e. , with randomized number of balls , see section [ sec2 ] below ) .
the prototypical ( apparently folklore ) instance of frequencies @xmath7 with converging variance
@xmath1 is the geometric sequence of ratio @xmath26 ( i.e. , @xmath27 ) , where one can show with some effort that @xmath28 as @xmath14 ( see @xcite ) .
note that our condition ( [ pv ] ) is obviously satisfied here with @xmath29 , hence the result .
the mechanism leading to such a simple answer is due to a resonance of the ratio @xmath30 with the intrinsic dyadic structure of the variance , resulting in massive cancelation of oscillating terms ( again , in the poissonized version , see example [ ex : geometric1/2 ] below ) .
recently , such cancelations have been explained directly for the original model ( i.e. , for @xmath1 ) using sophisticated analytic methods @xcite .
it seems to be less well known that for generic geometric frequencies @xmath31 , the ( finite ) limit of @xmath1 exists if @xmath32 ( @xmath33 ) , with the limiting value @xmath34 ( see ( * ? ? ? * , page 15 ) ) .
again , using theorem [ thm : limit ] one gets this answer immediately , together with the `` only if '' statement ; moreover , the same conclusion can be readily extended to sequences @xmath7 from the parametric class @xmath35 ( see @xcite ) , defined by the property @xmath36 thus asymptotically mimicking the geometric decay .
( some concrete examples of distributions in the @xmath35 class , complementing the geometric instance , will be given below in section [ sec1.3 ] . ) indeed , in the @xmath35 case equation ( [ pv ] ) amounts to @xmath37 , whence @xmath32 .
of course , condition ( [ eq : rt ] ) is too restrictive for the criterion ( [ pv ] ) , as can be seen for instance by merging @xmath23 geometric sequences of the same ratio @xmath30 ( and normalizing the resulting sequence so as to satisfy ( [ add1 ] ) ) .
the following `` decomposition '' interpretation of theorem [ thm : limit ] clarifies the compound structure of frequency sequences @xmath7 that exhibit convergence of the variance .
observe that by condition ( [ pv ] ) , the sequence @xmath7 splits in a disjoint fashion into @xmath23 non - increasing subsequences @xmath38 ( @xmath39 ) , each belonging to the @xmath40 class : @xmath41 moreover , by the `` if '' part of theorem [ thm : limit ] , each of the @xmath23 constituent subsequences brings a unit contribution to the overall limiting variance @xmath34 .
such a decomposition may be interpreted as splitting the initial array of boxes @xmath2 into @xmath23 infinite sub - arrays @xmath42 ( @xmath39 ) , and allocating the balls to boxes in a two - stage procedure as follows : for each ball , a destination array is chosen independently with probabilities @xmath43 , and the ball is then thrown with the corresponding ( re - scaled ) frequencies @xmath44 ( @xmath3 ) .
the additivity of the variance in this procedure , as predicted by theorem [ thm : limit ] , may be somewhat surprising , given the apparent dependence of the partial occupancy numbers @xmath45 ( @xmath39 ) .
however , additivity becomes quite transparent in the poissonized setting , where the dependence between boxes is removed ( see a remark in section [ sec2.2 ] ) .
historically , there has been some confusion about the converging variance in the geometric model .
controversy started in ( * ? ? ?
* example 6 ) , where karlin asserted that his sufficient condition for convergence ( * ? ? ?
* theorem 2 ) was satisfied for _ every _ geometric sequence @xmath46 ( @xmath47 ) , with the limiting value given by @xmath48 .
as we have seen , this is false unless @xmath49 belongs to the countable set @xmath50 .
a more careful inspection reveals that karlin s condition , if applied accurately , does yield the correct answer in the geometric case , properly discriminating between convergence vs. divergence !
moreover , we have found out , quite unexpectedly , that karlin s condition ( decorated in @xcite with some superfluous assumptions and originally conceived as just a sufficient condition ) proves to be _ necessary and sufficient _ , being equivalent to our own criterion proved in lemma [ lm : cesaro ] .
we will discuss this link below , in section [ sec5karlin ] . that there was something wrong with example 6 in @xcite was subsequently pointed out by dutko @xcite , who noticed that @xmath1 is bounded below by a positive constant , uniformly in @xmath0 and @xmath49 , hence the limit @xmath48 can not be valid at least for small values of @xmath49 ( when @xmath51 gets arbitrarily close to zero ) .
however , dutko @xcite apparently claimed that the limit of the variance fails to exist for _ each _
@xmath52 , thus missing the other values , @xmath32 , @xmath53 .
unfortunately , he gave no details to support such a conclusion , referring to his unpublished thesis @xcite , which is not easily available .
more recent studies @xcite have shed much light on the geometric model .
hitczenko and louchard @xcite ( motivated by random compositions of natural numbers ) were apparently first to prove analytically that @xmath54 in the geometric case with @xmath30 , contrary to `` popular belief '' @xcite that persistent oscillations are ubiquitous in discrete random structures involving geometric distribution ( see , e.g. , @xcite ) .
prodinger @xcite gave an alternative proof of this asymptotics ( along with a similar result for a particular model of data search trees called patricia tries ) , proceeding from the general `` oscillatory '' framework .
recently , archibald _
* theorem 2 ) derived a very precise asymptotic expansion @xmath55 where @xmath56 with @xmath57 periodic of period @xmath58 and zero mean ( the latter function emerges in a similar expansion for @xmath59 , the expected value of @xmath9 ) .
if @xmath30 then @xmath60 , and from the expansion ( [ arch ] ) it is seen that the oscillating term vanishes due to @xmath58-periodicity of @xmath57 , since @xmath61 ( see ( * ? ? ?
* appendix a , page 1079 ) .
in fact , the same argument is true for any @xmath32 ( @xmath33 ) , when @xmath62 and hence @xmath63 ( see ( * ? ? ? * , page 15 ) ) .
one can also wonder about conditions for other possible types of behavior of the variance @xmath1 .
we shall prove the following criterion of uniform boundedness , again set in terms of the lagged ratio @xmath64 compared to the _ upper _ threshold @xmath26 [ cf .
( [ pv ] ) ] .
[ thm : bound ] the sequence @xmath24 is bounded if and only if there exists a positive integer @xmath23 such that the frequencies @xmath7 satisfy the condition @xmath65 moreover if @xmath23 is the least integer with the property then @xmath24 satisfies a sharp asymptotic bound @xmath66 .
this situation is exemplified by the generic geometric frequencies , with arbitrary ratio @xmath47 .
another example is the poisson frequencies @xmath67 ( @xmath68 ) , where the variance @xmath1 is bounded but does not converge : indeed , here @xmath69 as @xmath70 , hence ( [ eq : boundsup ] ) is fulfilled whereas ( [ pv ] ) fails .
a larger class is that of quasi - binomial distributions @xcite , given by @xmath71 with parameters @xmath68 , @xmath72 .
( to explain the name , note that @xmath73 for @xmath74 , while for @xmath75 one has , in a continuous fashion , @xmath76 , thus recovering the poisson normalization constant . )
somewhat similar but different parametric family is given by the negative binomial distribution @xmath77 , with @xmath68 , @xmath47 [ here @xmath78 .
note that all these examples belong to classes @xmath35 with @xmath79 .
it is possible to construct more general examples using the `` decomposition '' reformulation of theorem [ thm : bound ] in the spirit of ( [ eq : decomposition ] ) , in that the variance @xmath1 is uniformly bounded if and only if the sequence @xmath7 may be split in a disjoint fashion into a finite number of subsequences , each of which satisfies condition ( [ eq : boundsup ] ) with @xmath29 ( e.g. , each from @xmath80 with @xmath81 , @xmath39 ) .
we shall also address the classical question of convergence to infinity and produce new conditions ensuring that @xmath13 .
note , however , that in contrast to the convergent or bounded cases , no necessary and sufficient criteria are available without extra regularity assumptions . to illustrate our results in this direction ,
let us formulate here two sufficient conditions , the first of which is set in terms of the lagged ratios @xmath64 against the _ lower _ threshold @xmath26 [ cf .
( [ eq : boundsup ] ) ] , while the second one is based on the `` tail ratio '' @xmath82 [ th:->infty ] suppose that for each integer @xmath20 @xmath83 then it follows that @xmath84 which in turn implies that @xmath13 as @xmath14 .
examples to theorem [ th:->infty ] are immediately supplied by the class @xmath85 , where condition ( [ eq:>= ] ) is obviously satisfied for any @xmath20 .
more complex examples ( not in @xmath85 ) will be constructed in sections [ sec4.1 ] and [ sec4.3 ] .
the tail ratio ( [ eq : rho ] ) can be expressed as @xmath86 , where @xmath87 is the discrete - time hazard rate , a key characteristic in reliability theory and survival analysis ( see , e.g. , @xcite ) .
the latter quantity also appears in the extreme value theory in connection with records from discrete distributions , where it is interpreted as the probability that @xmath5 is a record value ( see , e.g. , @xcite ) . in the occupancy context ,
condition ( [ eq : pjovertail0 ] ) is related to the `` probability of a tie for first place '' @xmath88 , where @xmath89 is the largest index among the occupied boxes after @xmath0 throws .
more specifically , it has been proved @xcite that condition ( [ eq : pjovertail0 ] ) is satisfied if and only if @xmath90 and moreover , if ( [ eq : pjovertail0 ] ) fails then @xmath88 does not converge at all .
this , combined with theorem [ sctail ] , shows that ( [ ptfp ] ) implies both @xmath13 and @xmath91 , which is a surprising connection between the behavior in the extreme - value range and the global characteristics of the sample .
these facts equally apply to the poissonized model .
the rest of the paper is organized as follows .
section [ sec2 ] contains general formulas and introduces the poissonization technique . in section [ sec3 ]
, we connect the variance @xmath1 with the mean number of singletons ( i.e. , the boxes occupied by exactly one of the first @xmath0 balls ) and derive useful upper bounds .
we also obtain here a basic integral representation of the poissonized variance @xmath92 via the laplace transform of the function @xmath93 , counting the frequencies @xmath4 in the interval @xmath94x/2,x]$ ] , and relate the threshold values of @xmath95 with the lagged ratios @xmath64 .
this analysis culminates in the proof of theorem [ thm : bound ] . in section [ sec4 ] , various sufficient conditions for @xmath13 are derived , which covers the content of theorem [ th:->infty ] .
we also show that these conditions are not necessary , by constructing examples of weird oscillatory behavior . in section [ sec5 ]
, we derive a simple integral condition in terms of the function @xmath95 , necessary and sufficient in order that @xmath92 converge to a finite limit .
this criterion is then used to prove theorem [ thm : limit ] . in conclusion ,
we rehabilitate karlin s sufficient condition of convergence , by showing that it is in fact necessary and sufficient .
let @xmath96 be the occupancy number of box @xmath5 after @xmath0 throws , that is , the number of balls out of the first @xmath0 that land in box @xmath5 .
note that @xmath97 where @xmath98 is the indicator of event @xmath99 ( i.e. , with values @xmath58 when @xmath99 is true and @xmath100 otherwise ) . because @xmath101 , it is clear that the terms in the sum ( [ knsum ] ) are not independent .
a common recipe to circumvent the dependence ( see @xcite for a general introduction and @xcite for details in the occupancy problem context ) is to consider a closely related model in which the balls are thrown at the jump times of a unit rate poisson process @xmath102 : by this randomization the balls appear in boxes according to independent poisson processes @xmath103 , with rate @xmath4 for box @xmath5 .
further advantage of the poissonized model is that the normalization can be replaced by a weaker summability condition @xmath104 , thus allowing one to avoid computing normalization constants in expressions for @xmath4 .
clearly , the normalization ( [ add1 ] ) can always be maintained by rescaling the frequencies @xmath105 , to the effect of a linear time change , @xmath106 . in what follows
, we adopt the convention that quantities derived from the poissonized version of the occupancy problem are written as functions of the continuous time parameter @xmath107 , while for the original model we preserve the notation with lower index @xmath0 . in particular , we write @xmath103 ( cf . above ) for the number of balls that land in box @xmath5 by time @xmath107 and @xmath108 for the number of boxes discovered by the poisson process @xmath109 .
likewise , denoting by @xmath110 the number of boxes , each of which is hit by exactly @xmath111 of the first @xmath0 balls , we write @xmath112 for the corresponding poissonized quantity ( which is the number of boxes containing exactly @xmath111 balls each by time @xmath107 ) . clearly ,
@xmath113 n&=\sum_r rk_{n,{\mbox{$\:\!$}}r}{\mbox{$\:\!$}},&\quad n(t)&= \sum_r
rk_r(t){\mbox{$\:\!$}}. \end{aligned}\ ] ] for the mean values of the number of occupied boxes we have the formulas @xmath114 related by the poissonization identity @xmath115 where @xmath116 . encoding the collection of frequencies into an infinite counting measure on @xmath1170,\infty[$ ] @xmath118 ( where @xmath119 is the dirac mass at @xmath120 , i.e. , @xmath121 for @xmath122 ) , we can represent the mean values ( [ eq : phin0 ] ) , ( [ eq : phit0 ] ) in an integral form as @xmath123 when the frequencies are normalized by ( [ add1 ] ) then all @xmath124 and the integral in ( [ eq : phiintt ] ) could be written in the limits from @xmath100 to @xmath58 , similarly to ( [ eq : phiintn ] ) . in the poissonized model , specific
normalization is not important , so we prefer to use a more flexible notation as in ( [ eq : phiintt ] ) .
the same convention applies to similar representations below ( see , e.g. , formulas ( [ eq : phitr ] ) and ( [ vardiff ] ) ) .
furthermore , set @xmath125 \label{eq : phitr } \phi_r(t)&:={\mathop{{}\mathbb e}{}}[k_{r}(t)]=\frac{t^r}{r ! } \int_0^\infty x^r { { \rm e}}^{-tx } \,\nu({{\rm d}}x){\mbox{$\:\!$}},\end{aligned}\ ] ] the latter being related to the derivatives of @xmath126 via @xmath127 note that equations ( [ eq : kandk ] ) imply @xmath128 n&=\sum_r r{\mbox{$\;\!$}}\phi_{n,{\mbox{$\:\!$}}r}{\mbox{$\:\!$}},&\quad t&= \sum_r r{\mbox{$\;\!$}}\phi_r(t){\mbox{$\:\!$}}. \end{aligned}\ ] ] an analyst will recognize in ( [ eq : phiintt ] ) a bernstein function ( see @xcite ) with the following general properties ( see also @xcite ) . [ bernstein ] if an infinite measure @xmath129 on @xmath130 satisfies @xmath131 then defines a function @xmath132 which 1 .
is analytic in the right half - plane 2 . has alternating derivatives @xmath133 @xmath134 3 .
satisfies @xmath135 but @xmath136 as @xmath137 . conversely if a function @xmath126 on @xmath138 has the properties and along with @xmath139 then there exists a unique infinite measure @xmath129 on @xmath130 such that representation holds . by the independence of summands in ( [ eq : defkt ] ) , the variance of @xmath140 is given by @xmath141 which is the same as @xmath142 [ ex : geometric1/2 ] for geometric frequencies of ratio @xmath30 , that is , @xmath27 ( @xmath3 ) , the sum ( [ eq : v ] ) is evaluated explicitly thanks to telescoping of partial sums ( see @xcite ) : @xmath143 in particular , it follows that @xmath144 as @xmath137 .
more generally , a similar simplification occurs in the geometric case with the ratio @xmath32 ( @xmath145 ) , where it is convenient to split the sum in ( [ eq : v ] ) into @xmath23 sub - sums ( over @xmath146 , where @xmath39 , @xmath147 ) , each involving a ( non - normalized ) geometric sequence with ratio @xmath26 . applying the previous result ( with @xmath30 ) and adding up the @xmath23 unit contributions emerging in the limit from the @xmath23 constituent subsequences
, we obtain the convergence @xmath148 as @xmath137 . for other values of @xmath49 the formula for the variance does not simplify .
the poissonized variance is additive : if @xmath149 and @xmath150 are two summable sequences of frequencies , and if @xmath7 is obtained by merging them into a single sequence , then the corresponding variances satisfy @xmath151 .
this explains the structural decomposition of the variance mentioned in the introduction and illustrated in example [ ex : geometric1/2 ] .
the fixed-@xmath0 counterpart of ( [ eq : v ] ) is @xmath152 where the cross - terms arise due to dependence in ( [ knsum ] ) . according to ( * ? ? ?
* proposition 4.3(ii ) ) , the variances @xmath153 and @xmath1 are always of the same order , @xmath154 in the next lemma , we establish estimates for the deviation of the poissonized quantities from their fixed-@xmath0 counterpart in terms of higher - order moments , which will be instrumental for depoissonization in the case of bounded variance ( see section [ sec3 ] ) .
[ varest ] if the normalization holds then @xmath155 \label{ee3 } v(n)-v_n&=o(n^{-1})\,\bigl(\phi_1(n)^2+\phi_2(n)\bigr){\mbox{$\:\!$}},\end{aligned}\ ] ] and for each @xmath156 @xmath157 we shall need the elementary inequalities @xmath158 the first inequality is obvious , while the second one follows from the estimate @xmath159 now , using representations ( [ eq : phiintn ] ) , ( [ eq : phiintt ] ) ( rewriting the integral ( [ eq : phiintt ] ) in the limits from @xmath100 to @xmath58 , due to ) and inserting the bounds ( [ eq:<exp < ] ) , we obtain @xmath160 which proves ( [ ee1 ] ) .
next , from ( [ eq : phinr ] ) and ( [ eq : phitr ] ) we get @xmath161 by the inequalities ( [ eq:<exp < ] ) , for each @xmath162 $ ] @xmath163 \label{eq : upper } { { \rm e}}^{-nx}-(1-x)^{n - r}&\le { { \rm e}}^{-nx}-(1-x)^{n}\le n{\mbox{$\:\!$}}x^2{\mbox{$\:\!$}}{{\rm e}}^{-nx}{\mbox{$\:\!$}}.\end{aligned}\ ] ] substituting the estimates ( [ eq : lower ] ) and ( [ eq : upper ] ) into ( [ eq : phi - phi ] ) and recalling the notation ( [ eq : phitr ] ) yields ( [ ee2 ] ) .
finally , as shown in ( * ? ? ?
* theorem 2.3 ) , the cross - terms in ( [ cross ] ) can be evaluated as @xmath164 & \quad+o\bigl(n^2 p_i^2 p_j^2{\mbox{$\;\!$}}(1-p_i)^{n-2}(1-p_j)^{n-2}\bigr){\mbox{$\:\!$}}.\end{aligned}\ ] ] inserting this estimate into ( [ cross ] ) and summing over all @xmath165 , we obtain @xmath166 from ( [ eq : phiandphi ] ) and ( [ eq : phiintn ] ) it follows that if the condition ( [ add1 ] ) holds then @xmath167 and similarly , using ( [ eq : phiintt ] ) , @xmath168 hence , subtracting ( [ eq : v_n_est ] ) from ( [ vardiff ] ) and using the estimates ( [ ee1 ] ) and ( [ ee2 ] ) , we arrive at ( [ ee3 ] ) .
in this section , we mainly focus on the situation where the variance @xmath92 is bounded .
we first derive various useful inequalities involving the functions @xmath92 , @xmath126 , @xmath169 and the measure @xmath129 .
since @xmath170 is decreasing and @xmath171 , the mean value theorem yields @xmath172 or equivalently @xmath173 the first inequality in ( [ phiv ] ) generalizes .
[ upbound ] for @xmath156 and @xmath174 @xmath175 setting @xmath176 ( see lemma [ bernstein](ii ) ) , we shall prove by induction the equivalent inequality @xmath177 suppose ( [ eq : phi_n ] ) holds for @xmath178 . note that @xmath179 , hence the function @xmath180 is convex and therefore @xmath181 on the other hand , since @xmath182 and by the induction hypothesis , @xmath183 combining ( [ eq : phi_r1 ] ) and ( [ eq : phi_r2 ] ) , we obtain ( [ eq : phi_n ] ) for @xmath184 .
thus , the induction step follows , and the proof is complete . consider the limits superior @xmath185 by continuity , @xmath92 is uniformly bounded on @xmath138 if and only if @xmath186 , and the same is true for @xmath169 in terms of the condition @xmath187 .
note that @xmath188 is strictly positive ( cf.@xcite ) ; indeed , setting @xmath189 in ( [ eq : v ] ) we have @xmath190 [ cor : v_phibar ] the conditions @xmath186 and @xmath191 are equivalent and imply @xmath187 for all @xmath192
. follows from ( [ phiv ] ) and lemma [ upbound ] .
appealing to lemma [ varest ] , we have depoissonization in terms of moments .
[ depoissmom ] if @xmath186 then as @xmath14 @xmath193 and for all @xmath194 @xmath195 lemma [ upbound ] entails an estimate of @xmath196 through either @xmath188 or @xmath197 . with some more effort
, we will derive an improved upper bound that does not depend on @xmath111 . recall that the measure @xmath129 is defined in ( [ eq : nu ] ) , and
consider the new ( finite ) measure @xmath198 when the normalization ( [ add1 ] ) holds , this is a probability measure governing the frequency distribution of the random box discovered by ball @xmath58 .
using the measure @xmath199 , we can rewrite ( [ eq : phitr ] ) as follows @xmath200 also , let us set @xmath201}{x}{\mbox{$\;\!$}}.\ ] ] [ lm : limsupphir ] suppose that @xmath186 .
then for all @xmath156 @xmath202 note that the last inequality in ( [ eq : phi_vbar ] ) follows from ( [ phiv ] ) .
further , integrating by parts in ( [ eq : phitr - nutilde ] ) and using the substitution @xmath203 , we get @xmath204\,{{\rm d}}y{\mbox{$\:\!$}}.\ ] ] for @xmath205 , due to monotonicity of the function @xmath206 $ ] , ( [ eq : phir ] ) implies @xmath207\,{{\rm d}}y \ge { { \rm e}}^{-1}{\mbox{$\;\!$}}\frac{\tilde{\nu}[0,1/t]}{1/t}{\mbox{$\;\!$}},\ ] ] and by letting here @xmath137 we obtain @xmath208 ( see ( [ eq : tilde_eta ] ) , ( [ eq : phi_vbar ] ) ) . on the other hand , for any @xmath192 from ( [ eq : phir ] ) it follows that @xmath209}{y / t}\,{{\rm d}}y\qquad ( r\ge 1){\mbox{$\:\!$}},\ ] ] which implies @xmath210 by the @xmath211 " part of fatou s lemma @xcite .
lemma [ lm : limsupphir ] implies that if @xmath186 then each term in the decomposition @xmath212 makes a uniformly bounded contribution to @xmath213 . this is to be contrasted with the case of frequencies akin to @xmath214 ( @xmath17 ) , where @xmath92 , @xmath126 and @xmath169 ( @xmath194 ) are of the same order @xmath215 as @xmath137 ( see @xcite ) .
the next lemma estimates the growth of @xmath126 in the case of bounded variance .
[ lm : phi < v ] suppose that @xmath186 .
then @xmath216 for any @xmath217 , there exists @xmath218 such that for all @xmath219 @xmath220 due to lemma [ lm : limsupphir ] .
therefore , @xmath221 hence , @xmath222 and since @xmath217 is arbitrary , our claim follows .
a shorter proof is by a simple `` @xmath211 '' version of lhpital s rule : @xmath223 due to lemma [ lm : limsupphir ] .
as in @xcite , it is convenient to rewrite the formula ( [ vardiff ] ) for the variance as a single integral representation .
recall that @xmath129 is given by ( [ eq : nu ] ) , and introduce the function @xmath224x/2,\,x]=\#\{j : x/2<p_j\le x\}\qquad(x>0){\mbox{$\:\!$}}.\ ] ] [ lm : v ] the variance @xmath92 can be represented as @xmath225 setting @xmath226x,\infty[$ ] and integrating by parts in ( [ vardiff ] ) gives @xmath227 and ( [ eq : v1 ] ) will follow if we show that @xmath228 as @xmath229 . to this end , note that the mean value of the measure @xmath129 is finite : @xmath230 .
hence , integration by parts yields @xmath231 and it follows that the limit in ( [ eq : by_parts ] ) exists and , moreover , must vanish , for otherwise the integral on the right - hand side of ( [ eq : by_parts ] ) would diverge .
[ cor : d ] the function @xmath232 is well defined and uniformly bounded for all @xmath233 . in particular @xmath234 .
letting @xmath235 in ( [ eq : v1 ] ) , we obtain @xmath236 hence @xmath237 for any @xmath238 .
vanishing at zero is obtained by the absolute continuity of the integral .
finally , boundedness of @xmath239 follows because @xmath240 for all @xmath120 large enough .
integrating by parts in ( [ eq : v1 ] ) and using corollary [ cor : d ] , we obtain an alternative representation , which will also be useful : @xmath241 it is immediately clear from ( [ eq : v1 ] ) that if @xmath242 for all @xmath238 then @xmath243 for all @xmath174 .
moreover , one can obtain two - sided asymptotic bounds as follows .
[ lm : bound ] recall that @xmath244 is given by and set @xmath245 then @xmath186 if and only if @xmath246 and in this case @xmath247 the substitution @xmath248 in ( [ eq : v1 ] ) yields @xmath249 and an application of the `` @xmath211 '' part of fatou s lemma @xcite implies @xmath250 for the converse inequality , we need to exploit the special structure of the measure @xmath129 . fixing @xmath238 and retaining in ( [ eq : v ] ) the terms with @xmath251x/2,x]$ ] only , we obtain @xmath252 } \bigl({{\rm e}}^{-tp}-{{\rm e}}^{-2tp}\bigr).\ ] ] it is clear that the minimum in ( [ eq : vn1 ] ) is attained at one of the endpoints , that is , @xmath253 or @xmath254 . setting @xmath255
$ ] , we note that @xmath256 y - y^2,&\phi\le y\le 1{\mbox{$\:\!$ } } , \end{array}\right.\ ] ] where @xmath257 is the golden ratio , which appears here as the root of the equation @xmath258 on @xmath940,1[$ ] .
it is then easy to see that the right - hand side of ( [ eq : vn1 ] ) , as a function of @xmath107 , attains its maximum value @xmath259 at @xmath260 ( @xmath261 ) .
hence @xmath262 , and the first inequality in ( [ eq : v < d < v ] ) follows .
our next goal is to characterize the link between the upper ( lower ) bounds on the values of the function @xmath93 ( for small @xmath120 ) and the lagged frequency ratios @xmath64 ( for large @xmath5 ) with regard to the threshold value @xmath26 .
[ p - and - nu ] for a given positive integer @xmath23 the bound @xmath263 is valid for all sufficiently small @xmath238 if and only if the condition @xmath264 is satisfied for all sufficiently large @xmath5 .
the similar assertion holds true when the sign @xmath265 in both and is replaced by @xmath266 .
the first part of the lemma ( i.e. , with @xmath265 ) is just a reformulation of definitions ( see ( [ eq : delta_nu ] ) ) . indeed , applying ( [ deltanuk ] ) with @xmath267 implies @xmath268 , which is ( [ eq : boundrat1 ] ) .
conversely , if @xmath269 then by ( [ eq : boundrat1 ] ) we have @xmath270 , and hence @xmath271x/2,x]\le k$ ] as required by ( [ deltanuk ] ) .
the `` mirror '' part ( i.e. , with @xmath266 ) needs a bit more care .
first , note that it suffices to prove the `` only if '' statement in the case where @xmath272 , for if @xmath273 ( @xmath274 ) then @xmath275 .
now , if @xmath276 then the condition @xmath277 implies that @xmath278 , whence by letting @xmath279 we get @xmath280 .
similarly , the `` if '' part follows by noting that @xmath280 implies @xmath277 for each @xmath276 . by lemma [ p - and - nu ] and the inequality ( [ eq : v < d < v ] ) , the upper bound ( [ deltanuk ] ) implies @xmath281 . in some cases , however ,
such an estimate may not be sharp , as the next example demonstrates .
[ ex : j/2 ] let @xmath282 , so by theorem [ thm : limit ] we have @xmath283 . on the other hand , ( [ eq : boundrat1 ] )
holds starting from @xmath284 , which leads to the crude bound @xmath285 .
an inspection shows that @xmath286 on @xmath287 and @xmath288 on @xmath289 ( @xmath290 ) .
for a given @xmath291 , `` excess '' over the value @xmath58 on the interval @xmath940,x]$ ] occurs on a set of total lebesgue s measure bounded by @xmath292 , which is small as compared to @xmath293 ( @xmath70 ) .
this example suggests the following refinement of lemma [ p - and - nu ] .
[ p - and - d ] if for some @xmath33 the frequencies @xmath7 satisfy @xmath294 then @xmath295 .
the assertion remains valid when the symbols @xmath265 and @xmath211 are simultaneously replaced by @xmath266 and @xmath296 .
it suffices to assume that @xmath29 , as the general case would then follow by the additivity argument ( see the remark after example [ ex : geometric1/2 ] ) . according to ( [ eq : boundrat11 ] ) ( with @xmath29 ) , for any @xmath2970,1/5]$ ] and all sufficiently large @xmath298 we have @xmath299 .
hence , @xmath300 , and lemma [ p - and - nu ] implies that @xmath301 for all sufficiently small @xmath120 . on the other hand , using the definition of the function @xmath95 one can check that @xmath302 when @xmath303 .
that is to say , the value @xmath304 may only occur on a subset of @xmath305 with lebesgue s measure not exceeding @xmath306 ( here @xmath307 , @xmath308 ) .
therefore , for @xmath309 we have @xmath310 inserting these estimates into ( [ eq : d ] ) , we obtain for @xmath311 @xmath312 it follows that @xmath313 hence @xmath314 . finally , applying to ( [ eq : v - d ] ) the `` @xmath211 '' part of fatou s lemma @xcite
, we obtain @xmath315 and the first half of the lemma is proved . for the second half ( with @xmath266 and @xmath296 ) , suppose again that @xmath29 . according to ( [ eq : boundrat11 ] ) , for any @xmath2970,1/2]$ ] and all sufficiently large @xmath298 , we have @xmath316 .
observe that possible deviations of the function @xmath95 from value @xmath58 may only occur as follows : if @xmath317 then @xmath318 for @xmath319 , while if @xmath320 then @xmath321 for @xmath322 .
in either case , the contribution of the interval with the endpoints @xmath323 and @xmath324 to the integral @xmath325 is bounded from below by @xmath326 using this remark , for a given @xmath327p_{j},p_{j-1}]$ ] we obtain @xmath328 hence @xmath329 and since @xmath330 is arbitrary , it follows that @xmath331 .
it remains to use fatou s lemma in ( [ eq : v - d ] ) to conclude that @xmath332 .
[ cor : lagged ] suppose that the condition is satisfied for some @xmath33 that is @xmath333 as @xmath70 .
then @xmath148 as @xmath137
. readily follows by combining the two halves of lemma [ p - and - d ] .
note that corollary [ cor : lagged ] is exactly the `` if '' part of theorem [ thm : limit ] . in section [ sec5 ]
below , where the issue of converging variance is considered in detail , we will give a direct , shorter proof of the necessity of the condition ( [ pv ] ) .
note that a converse statement to either half of lemma [ p - and - d ] is _ not valid_. indeed , if @xmath334 with @xmath335 , then @xmath286 on @xmath336 and @xmath337 on @xmath338 ( for @xmath298 large enough ) .
this implies that the graph @xmath339 consists of arcs of hyperbolas with alternating monotonicity ( supported on intervals of the form @xmath336 and @xmath338 ) , and in particular @xmath340}\frac{d(x)}{x}&=\frac{d(2p_j)}{2p_j } = \frac{1}{2p_j}\sum_{i\ge j } p_i=\frac{1+\rho_j}{2}{\mbox{$\:\!$}},\\ \min_{x\in[2p_{j},{\mbox{$\:\!$}}2p_{j-1}]}d(x)&=\frac{d(p_{j-1})}{p_{j-1}}=\frac{1}{p_{j-1}}\sum_{i\ge j } p_i = q{\mbox{$\:\!$}}(1+\rho_j){\mbox{$\:\!$ } } , \end{aligned}\ ] ] where @xmath341 ( cf .
( [ eq : rho ] ) ) .
the @xmath35-condition implies that @xmath342 as @xmath70 , so from ( [ eq : max / min ] ) we get @xmath343 in particular , setting @xmath75 ( e.g. , when @xmath7 is a poisson distribution ) and taking a `` doubled '' sequence ( i.e. , determined by @xmath344 ) , by the additivity argument we get @xmath345 , while @xmath346 . likewise , choosing @xmath347 and again doubling the sequence , from ( [ eq : dd ] ) and by fatou s lemma applied to ( [ eq : v - d ] )
, we obtain that @xmath348 , whereas @xmath349 .
we are now in a position to prove theorem [ thm : bound ] , and let us start by proving its poissonized version . by lemma [ lm : bound ] , the conditions @xmath186 and @xmath246 are equivalent , and according to the first half of lemma [ p - and - nu ] , the latter condition holds if and only if ( [ eq : boundrat1 ] ) is satisfied for some @xmath33 , which is equivalent to ( [ eq : boundsup ] ) ( possibly , with a bigger @xmath23 ) . the second part of the theorem ( leading to the estimate @xmath350 )
is settled by lemma [ p - and - d ] , since condition ( [ eq : boundrat11 ] ) of the lemma coincides with condition ( [ eq : boundsup ] ) of the theorem .
furthermore , by ( [ hj ] ) the condition @xmath351 is equivalent to @xmath186 , in which case also @xmath352 by corollary [ depoissmom ] .
finally , the optimality of the bound @xmath353 follows by merging @xmath23 geometric sequences with ratio @xmath30 each and using the additivity argument ( alternatively , one can consider the geometric frequencies with ratio @xmath32 ) .
thus , the proof of theorem [ thm : bound ] is complete .
let us remark that the threshold @xmath26 in theorem [ thm : bound ] is chosen to match neatly with theorem [ thm : limit ] .
replacing @xmath26 in ( [ eq : boundsup ] ) by some other value @xmath47 would lead to a more sophisticated upper bound @xmath354 where @xmath355 is the ceiling integer part of @xmath120 . indeed , iterating the condition @xmath356 , we get @xmath357 , provided that @xmath358 , and ( [ eq : c_k , q ] ) follows by lemma [ p - and - d ] . in fact , the constant @xmath359 here has the meaning of an upper bound for @xmath360 in the geometric case with ratio @xmath49 .
note that the representation ( [ arch ] ) leads to a similar ( in general , slightly better ) estimate @xmath361 ( cf .
( [ eq : c_k , q ] ) ) .
in this section , we establish new sufficient conditions in order that @xmath362 as @xmath137 ( which , in view of ( [ hj ] ) , is equivalent to @xmath13 as @xmath14 ) .
note that the combination of theorems [ suff1 ] and [ sctail ] ( to be proved in sections [ sec4.1 ] and [ sec4.2 ] , respectively ) along with discussion in section [ sec4.4 ] will settle theorem [ th:->infty ] stated in the introduction .
it is natural to seek a condition for @xmath362 based on the representation ( [ eq : v ] ) , that is , in terms of the function @xmath93 . in turn
, such a condition may be transformed into the information about the lagged ratio @xmath64 ( cf .
theorem [ thm : bound ] ) .
[ suff1 ] the condition @xmath363{\lim_{x\downarrow 0}\delta\nu(x)=\infty}\ ] ] implies that @xmath364{\forall{\mbox{$\:\!$}}k\in{{\mathbb n}}{\mbox{$\:\!$}},\ \ \quad \liminf_{j\to\infty } \frac{p_{j+k}}{p_j}\ge \frac{1}{2}{\mbox{$\;\!$}},}\ ] ] which in turn implies that @xmath362 as @xmath137 .
if condition ( [ eq:(i ) ] ) holds then for any @xmath33 we have @xmath277 for all sufficiently small @xmath238 . by lemma
[ p - and - nu ] , this implies that @xmath365 for all @xmath5 large enough , and ( [ eq:(ii ) ] ) follows .
further , condition ( [ eq:(ii ) ] ) implies convergence of @xmath92 to infinity by lemma [ p - and - d ] .
note that condition ( [ eq:(ii ) ] ) is obviously fulfilled for any sequence @xmath7 from @xmath85 , in which case it is well known that @xmath362 @xcite .
the next example demonstrates that there are instances of frequencies @xmath7 satisfying ( [ eq:(i ) ] ) but _ not _ in @xmath85 .
this example will also show that conditions ( [ eq:(i ) ] ) and ( [ eq:(ii ) ] ) of theorem [ suff1 ] are _ not necessary _ in order that @xmath362 .
[ ex:3^k ] let @xmath47 and suppose that the sequence @xmath7 consists of the values @xmath366 , each repeated @xmath298 times ( @xmath367 ) , which corresponds to the measure @xmath368 . note that the sequence @xmath7 is not in any @xmath369-class , since @xmath370 but @xmath371 .
however , for any @xmath3720,1[$ ] we have @xmath362 , since for @xmath373 $ ] @xmath374}\bigl({{\rm e}}^{-y}-{{\rm e}}^{-2y}\bigr)=j{\mbox{$\;\!$}}\bigl({{\rm e}}^{-1/q}-{{\rm e}}^{-2/q}\bigr ) \to \infty\qquad(j\to\infty){\mbox{$\:\!$}}.\end{aligned}\ ] ] if @xmath375 then for @xmath376 we have @xmath377 as @xmath261 , and condition ( [ eq:(i ) ] ) is valid . on the other hand , if @xmath378 then @xmath379 for @xmath380 , hence @xmath381 and ( [ eq:(i ) ] ) fails . also , for any @xmath382 , we have @xmath383 , so condition ( [ eq:(ii ) ] ) is not valid .
a different sufficient condition exploits the link between @xmath92 and the mean number of singleton boxes @xmath384 , as in lemma [ lm : limsupphir ] .
an equivalent condition may be set in terms of the tail ratio @xmath385 ( see ( [ eq : rho ] ) ) .
recall the definition ( [ eq : defnut ] ) of the measure @xmath199 .
[ sctail ] the condition @xmath386{\lim_{x\downarrow 0 } \frac{\tilde{\nu}[0,x]}{x}=\infty}\ ] ] is equivalent to @xmath387{\lim_{j\to\infty } \rho_j=\infty{\mbox{$\:\!$}},}\ ] ] and each one implies that @xmath362 as @xmath137 . by the estimate ( [ eq : tilde_rho ] ) , condition ( [ eq : nutoverx ] ) implies @xmath388 , which is equivalent to @xmath362 by .
so it remains to show that ( [ eq : nutoverx ] ) and ( [ eq : pjovertail ] ) are equivalent to each other .
observe that for @xmath389 we have @xmath390\ge\rho_j$ ] , hence ( [ eq : pjovertail ] ) implies ( [ eq : nutoverx ] ) . to prove the converse , note that if @xmath391 then @xmath392 , so it suffices to consider the case where @xmath393
. then @xmath394}{x}\to\infty\qquad(j\to\infty){\mbox{$\:\!$}},\ ] ] when the condition ( [ eq : nutoverx ] ) holds , and hence ( [ eq : pjovertail ] ) follows .
we construct here an example demonstrating that conditions ( [ eq : nutoverx ] ) , ( [ eq : pjovertail ] ) are _ not necessary _ in order that @xmath362 ( or , equivalently , @xmath388 ) .
in particular , due to the estimate ( [ eq : phi_vbar ] ) ( with @xmath395 ) , this example will show that @xmath362 does not necessarily imply @xmath396 . on the other hand , in view of the inequality @xmath3971/(2t),1/t]\bigr\ }
= \frac{{{\rm e}}^{-1}}{4}\,\delta\nu ( 1/t){\mbox{$\:\!$}},\end{aligned}\ ] ] it is a priori clear that @xmath398 can not be uniformly bounded in such a situation , because @xmath399 according to ( [ eq : v < d < v ] ) .
[ ex : phi1notphi2 ] let @xmath400 be an increasing integer sequence .
take the frequencies @xmath7 in the form @xmath401 \displaystyle k_{i+1}^{-1},\ \
& k_0+\dots+k_{i-1}<j\le k_0+\dots+k_i{\mbox{$\;\!$ } } , \end{array } \right.\ ] ] which corresponds to the measure @xmath402 that is to say , the array of boxes is partitioned in blocks so that @xmath298-th block contains @xmath403 boxes of frequencies @xmath404 ( @xmath405 ) .
the heuristics underlying this example is as follows .
a prototype instance is a block of @xmath23 equal boxes each with frequency , say , @xmath49 .
the mean number of singleton boxes within the block is a single - wave function @xmath406 which increases to its maximum @xmath407 at time @xmath408 and then goes down to @xmath100 .
now , the idea is to combine a series of such blocks in order to guarantee a suitable overlap of the waves produced by successive blocks .
if the sequence @xmath409 grows fast enough , then for each @xmath405 there exists a time instant ( of order of @xmath410 ) when boxes belonging to @xmath298-th block start to get occupied . after some time
, the mean number of singletons among these boxes is still relatively large , say not less than @xmath411 , but the expected number of balls that fall in boxes of further blocks becomes large too , and almost all these balls produce singleton boxes , since @xmath410 is yet much larger ( hence the frequencies are smaller ) . as time passes , all boxes belonging to blocks @xmath412 are likely to contain more than one ball each , while the balls hitting other blocks remain sole representatives of their boxes . to make this heuristic work , we choose @xmath413 so that @xmath414 for all @xmath298 .
we wish to check that @xmath384 goes to infinity but @xmath398 does not . using ( [ eq : phitr ] ) and ( [ eq : tnu ] ) we have @xmath415 as a function of @xmath107 , each summand @xmath416 in the sum ( [ eq : exphi1 ] ) increases up to the maximum value @xmath417 attained at @xmath418 , and then decreases to zero .
two consecutive summands , @xmath416 and @xmath419 , are equal at the point @xmath420 where their common value is @xmath421 using the elementary inequality @xmath422 ( @xmath53 ) , we note that @xmath423 since @xmath424 @xmath425 , it follows that for all @xmath426 $ ] , @xmath427 hence @xmath428 turning to @xmath398 , note that the summand @xmath429 in ( [ eq : exphi2 ] ) attains its maximum value at the point @xmath430 and @xmath431 , so @xmath432 on the other hand , on the sequence @xmath433 one has @xmath434 setting @xmath435 and @xmath436 , we note that the function @xmath437 increases for @xmath438 .
hence , for all @xmath439 , @xmath440 and therefore @xmath441 for @xmath442 , we have @xmath443 and since @xmath444 for @xmath290 , it follows @xmath445 combining the estimates ( [ eq : j1 ] ) and ( [ eq : j2 ] ) yields @xmath446 as @xmath70 . thus @xmath398 does not have a limit as @xmath137 , and moreover @xmath447 finally , it is easy to see directly that in this example the limit in ( [ eq : pjovertail ] ) does not exist .
indeed , along the subsequence @xmath448 , according to ( [ eq : p_ex ] ) and ( [ eq : k_i ] ) , @xmath449 on the other hand , for @xmath450 we have @xmath451 karlin @xcite gives an example of frequencies for which @xmath92 converges to @xmath100 along a sequence of values of @xmath107 , and converges to @xmath452 along another sequence ; in that case @xmath384 demonstrates the same type of behavior .
our example [ ex : phi1notphi2 ] exhibits a more exotic `` second order '' pathology : this time , @xmath453 but @xmath398 oscillates between @xmath100 and @xmath452 .
first of all , note that condition ( [ eq:(ii ) ] ) in theorem [ suff1 ] does not imply condition ( [ eq:(i ) ] ) .
a counterexample may be constructed by a slight modification of example [ ex:3^k ] as follows : define the frequencies @xmath7 by setting @xmath454 , where @xmath455 , then @xmath456 ( so that ( [ eq:(ii ) ] ) is satisfied ) , but for @xmath457 we have @xmath379 , hence @xmath458 and ( [ eq:(i ) ] ) fails .
further , it is easy to see that condition ( [ eq:(ii ) ] ) in theorem [ suff1 ] implies the set of equivalent conditions ( [ eq : nutoverx ] ) , ( [ eq : pjovertail ] ) in theorem [ sctail ] , but not the other way around .
indeed , if ( [ eq:(ii ) ] ) is satisfied then for @xmath459 defined in ( [ eq : rho ] ) we have @xmath460 and condition ( [ eq : pjovertail ] ) follows .
on the other hand , we have seen that in example [ ex:3^k ] condition ( [ eq:(ii ) ] ) fails , while for @xmath461 we have @xmath462}{x}=\frac{1}{x}\sum_{i\ge j}^\infty i{\mbox{$\:\!$}}q^{i}\ge \frac{j}{q^{j-1}}\sum_{i\ge j}^\infty q^{i}=\frac{j{\mbox{$\:\!$}}q}{1-q}\to \infty\qquad(j\to\infty){\mbox{$\:\!$}},\ ] ] and the condition ( [ eq : nutoverx ] ) is valid .
as example [ ex : phi1notphi2 ] shows , a converse to theorem [ sctail ] is _ not valid _ , unless under further assumptions on the measure @xmath199 ( cf .
for instance , if @xmath463 $ ] varies regularly at zero , then karamata s tauberian theorem ( see @xcite or @xcite ) applied to ( [ eq : phitr - nutilde ] ) yields @xmath463/x\sim c\,{\phi_1(1/x)}$ ] as @xmath261 , so that the convergence @xmath388 as @xmath137 does imply the condition ( [ eq : nutoverx ] ) . by karamata
s tauberian theorem , the convergence @xmath464 is equivalent to @xmath463/x \to c$ ] as @xmath261 .
interestingly , the implication may fail for @xmath465 , as example [ ex : phi1notphi2 ] demonstrates .
we will now investigate the situation where the variance @xmath92 has a finite limit as @xmath137 , which is the central topic of this work ( see theorem [ thm : limit ] ) . as already mentioned in section [ sec3.6 ] , the `` if '' part of theorem [ thm : limit ] follows from corollary [ cor : lagged ] .
so the main goal of this section is to prove the `` only if '' part ( i.e. , the sufficiency of the condition ( [ pv ] ) ) , but we will also give a streamlined proof of the necessity . recall that @xmath466 is a primitive function of @xmath95 , defined by ( [ eq : d ] ) .
[ lm : cesaro ] in order that there exist a finite limit @xmath467 it is necessary and sufficient that @xmath468 note that , according to ( [ eq : vbar>0 ] ) , @xmath469 . by the representation ( [ eq : v1 ] ) , we can rewrite ( [ eq : var_limit ] ) as @xmath470 by karamata s tauberian theorem ( see @xcite , @xcite ) , the relation ( [ eq : d_n ] ) is equivalent to @xmath471 as @xmath261 , which is the same as ( [ eq : cesaro ] ) .
[ lm : alphabeta ] suppose that the limit exists and let @xmath472 be arbitrary variables such that @xmath473 and @xmath474 .
then @xmath475 using ( [ eq : cesaro ] ) , we have @xmath476 since the ratio @xmath477 is bounded .
[ lm : integer ] if the finite limit ( [ eq : var_limit ] ) exists then the limiting value @xmath22 must be a positive integer number @xmath478 and in this case @xmath4790,x]:\delta\nu(u)\ne k\}}{x}=0{\mbox{$\:\!$}},\ ] ] where @xmath480 denotes lebesgue s measure on @xmath130 . by lemma [ lm : bound ] , the function @xmath481 is uniformly bounded . by definition , it counts the number of frequencies @xmath4 in the interval @xmath94u/2,u]$ ] , therefore @xmath481 is piecewise constant , with jumps at points @xmath482 and @xmath483 .
thus , for any given interval @xmath94x/2,x]$ ] the total number of such jumps is uniformly bounded by a constant , say @xmath484 .
let @xmath94\alpha,\beta[$ ] be the maximal open subinterval of @xmath94x/2,x]$ ] , on which @xmath95 is constant .
clearly , its length satisfies @xmath485 , thus @xmath486 consider a closed interval @xmath487\subset { } ] \alpha,\beta[$ ] with @xmath488 , @xmath489 . since @xmath490 and @xmath491 , by the bound ( [ eq : m ] ) lemma [ lm : alphabeta ] applies to yield @xmath492 but the function @xmath95 is constant on @xmath94\alpha,\beta[{}\supset[\alpha_1,\beta_1]$ ] , hence its sole ( integer ) value must coincide with the asymptotic mean @xmath22 given by ( [ eq : mean ] ) .
in particular , @xmath22 must be integer , @xmath478 . along the same lines , one can show that for any @xmath217 and all small enough @xmath120 , the function @xmath95 takes the value @xmath34 on the interval @xmath94x/2,x]$ ] everywhere except on a set of lebesgue s measure smaller than @xmath493 .
thus , lebesgue s measure of the set @xmath4940,x]:\delta\nu(u)\ne k\}$ ] is bounded by @xmath495 , and since @xmath496 is arbitrary , ( [ eq : lambda ] ) follows . [
th : pseudo ] if the limit exists hence @xmath478 by lemma then cf .
@xmath497 without loss of generality , it suffices to consider @xmath498 such that @xmath499 .
suppose first that @xmath500 .
then for @xmath501 we have @xmath94x/2,x]\subset { } ] p_{j+k},p_j[$ ] and hence @xmath502 . therefore ,
@xmath503 using that @xmath504 as @xmath261 ( see lemma [ lm : cesaro ] ) , from ( [ eq : pk ] ) we deduce that @xmath505 , which , together with the hypothesis @xmath506 ( see above ) , implies ( [ eq : pk / pkv ] ) . likewise , if @xmath507 then for @xmath508 we have @xmath94x/2,x]\supset [ p_{j+k},p_j]$ ] , hence @xmath509 and ( cf .
( [ eq : pk ] ) ) @xmath510 similarly as before , this simplifies to @xmath511 , and since we assumed that @xmath506 , ( [ eq : pk / pkv ] ) follows .
the proof is complete .
let us now show the converse of lemma [ th : pseudo ] ( as mentioned at the beginning of section [ sec5 ] , this also follows from corollary [ cor : lagged ] ) .
[ th : pseudo1 ] assume that the sequence @xmath7 satisfies the condition for some @xmath33 .
then the limit exists and @xmath34 .
by additivity , it suffices to prove that for each subsequence @xmath512 ( @xmath513 ) , its contribution to the limit ( [ eq : var_limit ] ) equals exactly @xmath58 .
thus the proof is reduced to showing that if @xmath514 then @xmath515 by the @xmath369-condition , @xmath516 , where @xmath517 as @xmath70 .
hence , for any @xmath2970,1/3]$ ] and all @xmath5
large enough we have @xmath518 .
in particular , @xmath519 , which implies by lemma [ p - and - nu ] that @xmath520 for small @xmath120 . by lemma
[ lm : bound ] and the estimate ( [ phiv ] ) , it follows that @xmath521 is bounded .
returning to ( [ eq:=1 ] ) , observe that @xmath522 by the inequality @xmath523 , the sum in ( [ eq : gamma_k ] ) is dominated by @xmath524 passing to the limit in ( [ eq : gamma_k ] ) as @xmath525 , we obtain @xmath526 as @xmath137 , and since @xmath496 is arbitrarily small , we arrive at ( [ eq:=1 ] ) .
we are now able to complete the proof of our main theorem [ thm : limit ] characterizing the case of converging variance .
indeed , putting together lemmas [ th : pseudo ] and [ th : pseudo1 ] yields the desired criterion for @xmath527 . appealing to corollary [ depoissmom ] we conclude that the same condition applies to @xmath528 . in conclusion ,
let us recall that karlin s sufficient condition for @xmath527 ( * ? ? ?
* theorem 2 ) involves ( i ) the condition @xmath529 and ( ii ) an integral condition , which in our notation reads @xmath530 or , after an obvious change of variables , @xmath531 throughout his paper , karlin also postulates that the function @xmath532x,\infty[$ ] is regularly varying at zero ( see @xcite .
as we shall see , this condition is superfluous and may be omitted ( in fact , karlin s proof of his theorem 2 only requires the boundedness of @xmath93 , which follows easily from condition ( i ) ) .
note that condition ( i ) itself is not necessary for the convergence of @xmath92 : for instance , it does not hold for a sequence @xmath7 obtained by merging several geometric sequences with ratio @xmath26 into one .
furthermore , application of condition ( [ eq : karlin0 ] ) to the geometric case ( with ratio @xmath49 ) yields the following ( cf .
* example 6 ) containing an error ) .
let @xmath533 , where @xmath534 $ ] is the integer part of @xmath535 and @xmath536 is its fractional part .
from the definition of @xmath95 it follows that @xmath537-[\log_{1/q}y]\right){{\rm d}}y\\ \notag&= \frac{1}{x}\int_0^x \left([k+\delta + \log_{1/q } y]-[\log_{1/q}y]\right){{\rm d}}y\\ \label{eq : k+}&= k+\frac{1}{x}\int_0^x \left([\delta+\log_{1/q}y]-[\log_{1/q}y]\right){{\rm d}}y{\mbox{$\:\!$}}.\end{aligned}\ ] ] if @xmath538 , the integral in ( [ eq : k+ ] ) vanishes and condition ( [ eq : karlin0 ] ) yields @xmath34 .
however , if @xmath539 then ( [ eq : k+ ] ) does not have a limit as @xmath540 , since for @xmath541 the integral term amounts to @xmath542 whereas for @xmath543 it reads @xmath544 as a result , condition ( [ eq : karlin0 ] ) is satisfied if and only if @xmath545 , or equivalently @xmath32 .
our theorem [ thm : limit ] gives the same result , so ( [ eq : karlin0 ] ) proves to yield a correct answer in the whole range of the geometric case .
this observation brings up the question about the exact relationship between karlin s condition ( [ eq : karlin0 ] ) ( or ( [ eq : karlin ] ) ) and our criterion ( [ eq : cesaro ] ) .
surprisingly enough , we can demonstrate the following .
[ th : karlin ] condition is equivalent to and hence the former is necessary and sufficient in order that @xmath527 as @xmath137 .
suppose condition ( [ eq : cesaro ] ) holds . using the notation @xmath239 ( see ( [ eq : d ] ) ) and integrating by parts , we get @xmath546 where we used that the function @xmath547 is bounded on @xmath940,\infty[$ ] ( in particular , the dominated convergence theorem can be applied ) . hence , ( [ eq : karlin ] ) follows . on the other hand , condition ( [ eq : karlin ] )
amounts to @xmath548 again integrating by parts , we obtain @xmath549 where we may use dominated convergence because the function @xmath550 is bounded on @xmath940,1]$ ] due to ( [ eq : xg ] ) .
thus , condition ( [ eq : karlin ] ) implies ( [ eq : cesaro ] ) , and the proof is complete .
the statement of theorem [ th : karlin ] is a particular case of a general karamata theorem ( see @xcite , @xcite ) , according to which the limiting relation ( [ eq : cesaro ] ) is equivalent to either of the limits @xmath551 ( note that ( [ eq : cesaro ] ) itself is contained in the second formula with @xmath552 . ) that is to say , our condition ( [ eq : cesaro ] ) may be included in a parametric family of mutually equivalent criteria , set in terms of rescaled integrals of the function @xmath95 against polynomial weights ( the canonical criterion ( [ eq : cesaro ] ) being apparently the simplest ) .
we have given a direct proof of theorem [ th : karlin ] because of the historic interest of karlin s condition ( [ eq : karlin0 ] ) .
main part of this research was carried out during l. v. bogachev s visit to the university of utrecht in june 2006 , made possible due to a grant under the global exchange programme ( gep ) of the world universities network ( wun ) , awarded at the university of leeds .
hospitality of the hosts at utrecht is much appreciated .
archibald , m. , knopfmacher , a. and prodinger , h. ( 2006 ) . the number of distinct values in a geometrically distributed sample .
_ european j. combin . _
* 27 * 10591081 .
bahadur , r. r. ( 1960 ) . on the number of distinct values in a large sample from an infinite discrete distribution .
india part a _ * 26 * ( suppl .
ii ) 6775 .
baryshnikov , yu . ,
eisenberg , b. and stengle , g. ( 1995 ) . a necessary and sufficient condition for the existence of the limiting probability of a tie for first place .
_ statist .
probab.lett._ * 23 * 203209 . darling , d. a. ( 1967 ) .
some limit theorems associated with multinomial trials . in _ proc .
fifth berkeley sympos .
math.statist .
berkeley , ca , 1965/66 _ * ii * _ contributions to probability theory part 1 _ ( l. m. le cam and j. neyman , eds ) 345350 .
california press , berkeley , ca .
gnedin , a. and pitman , j. ( 2006 ) .
moments of convex distribution functions and completely alternating sequences . in _
festschrift for avner friedman ims lecture notes monogr .
statist . ,
beachwood , oh ( to appear ) .
available at ` arxiv : math.pr/0602091 ` .
ivanov , v. a. , ivchenko , g. i. and medvedev , yu .
i. ( 1984 ) .
discrete problems in probability theory .
( russian ) in _ itogi nauki tekhn .
veroyatnost .
. teor .
kibernet . _
* 22 * 360 .
vsesoyuz.inst . nauchn .
i tekhn .
( viniti ) , moscow .
[ english translation : ( 1985 )
. _ j. soviet math . _ * 31 * 27592795 . ]
kerov , s. v. ( 1995 ) .
coherent allocations , and the ewens - pitman formula .
( russian ) preprint , pomi 21/1995 , 115 .
[ english translation : ( 2005 ) . coherent random allocations , and the ewens - pitman formula .
( edited and with comments by a. gnedin . ) in _ representation theory dynamical systems combinatorial and algorithmic methods part 12 _ ( a. m. vershik , ed . ) , _ zap.nauchn .
s .- peterburg .
otdel . mat .
steklov.pomi_ * 325 * 127145 . ] kolchin , v. f. , sevastyanov , b. a. and chistyakov , v. p. ( 1976 ) .
_ random allocations_. ( russian ) izdat .
`` nauka '' , moscow .
[ english translation : ( 1978 ) .
v. h. winston & sons , washington , d.c . ; halsted press ( john wiley & sons ) , new york .
] , kotz , s. and balakrishnan , n. ( 1997 ) .
advances in urn models during the past two decades . in _ advances in combinatorial methods and applications to probability and statistics _ ( n. balakrishnan , ed . ) 203257 .
birkhuser , boston .
louchard , g. , prodinger , h. and ward , m. d. ( 2005 ) .
the number of distinct values of some multiplicity in sequences of geometrically distributed random variables . in _
2005 international conference on analysis of algorithms _ , discrete math .
sci.proc . , ad ( c. martnez , ed . ) 231256 ( electronic ) .
dmtcs , nancy .
available at ` http://www.dmtcs.org/pdfpapers/dmad0122.pdf ` .
nevzorov , v. b. ( 2000 ) .
_ records mathematical theory_. ( russian ) izdat .
fazis , moscow .
[ english translation : , prodinger , h. ( 2004 ) .
compositions and patricia tries : no fluctuations in the variance ! in _ proceedings of the sixth workshop on alenex and the first workshop on analco , new orleans , 2004 _ ( l. arge , g. italiano and r. sedgewick , eds ) 211215 .
siam , philadelphia , pa . available at ` http://math.sun.ac.za/~prodinger/ ` ` pdffiles / new_orleans.pdf ` .
sedgewick , r. and flajolet , p. ( 1996 ) .
_ an introduction to the analysis of algorithms_. addison - wesley , boston .
szpankowski , w. ( 2001 ) .
_ average case analysis of algorithms on sequences_. wiley - interscience , new york . | we consider the occupancy problem where balls are thrown independently at infinitely many boxes with fixed positive frequencies .
it is well known that the random number of boxes occupied by the first @xmath0 balls is asymptotically normal if its variance @xmath1 tends to infinity . in this work
, we mainly focus on the opposite case where @xmath1 is bounded , and derive a simple necessary and sufficient condition for convergence of @xmath1 to a finite limit , thus settling a long - standing question raised by karlin in the seminal paper of 1967 .
one striking consequence of our result is that the possible limit may only be a positive integer number .
some new conditions for other types of behavior of the variance , like boundedness or convergence to infinity , are also obtained .
the proofs are based on the poissonization techniques . |
according to cosmological n - body / hydrodynamic simulations , intergalactic shock waves develop as a consequence of the large scale structure formation of the universe .
infall of baryonic gas toward sheets , filaments and knots , as well as supersonic flows associated with hierarchical clustering , induce shocks ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . those cosmological shock waves , like most astrophysical shocks , are `` collisionless '' features mediated by collective , electromagnetic viscosities . such viscosities rely on irregular magnetic field components , mhd wave turbulence that is self - excited by the streaming suprathermal particles produced during shock formation ( see , e.g. , * ?
* ; * ? ? ?
* ; * ? ? ?
the existence and character of these shocks is important for several reasons . through dissipation
the cosmological shock waves convert part of the gravitational energy associated with structure formation into heat and , thus , govern the gas thermal history in the universe .
thermal energy is then radiated away , manifesting the large scale structure as well as the dynamics that created it ( see , e.g. , * ? ? ?
* ; * ? ? ?
at the same time , due to incomplete plasma thermalization at collisionless shocks , a sizeable portion of the shock energy can be converted into cosmic ray ( cr ) energy ( mostly ionic ) via diffusive shock acceleration ( dsa ) ( for reviews , see , e.g. , * ? ? ? * ; * ? ? ? * ) .
this nonthermal population represents a small fraction of the ion flux through a collisionless shock that leaks back upstream to become subject to dsa ; that is , to be `` injected '' into the nonthermal population . numerous nonlinear studies of dsa have shown that substantial fractions of the energy flux through strong shocks can be captured by the nonthermal populations ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
extensive nonlinear simulations by some of us incorporating a plasma - physics - based `` thermal leakage '' injection model into combined gas dynamic / cr diffusion - convection simulations found that strong shocks transfer up to @xmath7 of the initial shock kinetic energy to crs by this process @xcite .
there is clear evidence that one or more processes energize significant nonthermal particle populations in and around cosmic structures .
a number of clusters have been found with diffuse synchrotron radio halos or / and radio relic sources , indicating the existence of relativistic electron populations in intracluster medium ( icm ) ( see , e.g. , * ? ? ?
* ; * ? ? ?
in addition , some clusters have been reported to possess excess euv and/or hard x - ray radiation compared to that expected from the hot , thermal x - ray emitting icm , most likely produced by inverse compton scattering of cosmic microwave background radiation ( cmbr ) photons by cr electrons ( see , e.g. , * ? ? ? * ; * ? ? ? * ) . also it has been suggested that a fraction of the diffuse @xmath9-ray background radiation could originate from the same process @xcite .
if some of those cr electrons have been energized at cosmological shock waves , the same process should have produced greater cr proton populations .
hence , although cr protons in the icm have yet to be confirmed by the observation of @xmath9-ray photons produced by inelastic collisions between cr and thermal - gas protons ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ) , there may very well exist cr proton populations there whose pressure is comparable to the gas thermal pressure @xcite .
the properties of cosmological shock waves in the large scale structure of the universe were analyzed quantitatively by @xcite using numerical simulations with @xmath10 grid zones in a cubic comoving region of size @xmath11 mpc for a scdm model and a @xmath0cdm model .
they identified accretion shocks , merger shocks and internal flow shocks which were formed by infall and hierarchical clustering , and showed that the topology of these shocks is very complex .
they found merger and internal flow shocks distributed over mach numbers from 3 to 10 with a peak at @xmath12 and accretion shock mach numbers ranging between 10 and a few @xmath13 . in a recent study based on a higher resolution simulation with @xmath14 grid zones in a cubic comoving region of size @xmath15 mpc for a @xmath0cdm model
, @xcite showed that most dissipation involves shocks of modest strength , with @xmath16 accounting for @xmath17 of total shock heating . on the other hand , through a semi - analytic study @xcite found a mach number distribution of merger - related shocks during large scale structure formation with a peak at much lower mach number ( @xmath18 ) . in this paper
, we critically re - examine the properties of cosmological shock waves with a new set of cosmic structure formation simulations .
we quantify the characteristics of cosmological shock waves and estimate the dissipation , gas thermalization and cr acceleration at those shocks .
the capture of shocks in hydrodynamic simulations and the identification of shocks in such simulation data are affected by numerical details including resolution .
therefore , to validate our findings we estimate the errors in our measured quantities through consistency checks and resolution convergence tests .
this study adds valuable insights into the thermal history and nature of gas in the universe , as well as nonthermal activities in the icm . in 2 ,
simulations are detailed along with shock identification . the main results of shock characteristics and shock dissipation are described in 3 and 4 , respectively , followed by a summary in 5 .
the cold dark matter cosmology with a cosmological constant ( @xmath0cdm ) was employed with the following parameters : @xmath19 , @xmath20 , and @xmath21 , @xmath22/(100 km / s / mpc ) = 0.7 , and @xmath23 .
the above values are consistent with those fitted with the recent wmap data ( see , e.g. , * ? ? ?
. a cubic region of comoving size @xmath24 mpc was simulated inside a computational box with @xmath25 , @xmath14 , @xmath26 , @xmath27 and @xmath28 grid zones for gas and gravity and @xmath14 , @xmath26 , @xmath27 , @xmath28 and @xmath29 particles for dark matter .
it allows a uniform spatial resolution of @xmath30 kpc @xmath31 @xmath32 mpc .
the simulations were performed using a pm / eulerian hydrodynamic cosmology code .
the code is described in @xcite .
but the version of the code used includes several updates . for instance
, it now adopts the mc ( monotonized central difference ) limiter , instead of the original minmod limiter .
the update to the mc limiter was intended to enhance the density resolution and to capture shocks more sharply ( see , e.g. , * ? ? ? * for details ) .
we did not include in our simulations several physical processes such as radiative cooling , galaxy / star formation , feedback from galaxies and stars , that can play significant roles in determining conditions within cluster cores , nor reionization of the intergalactic medium that effectively sets a temperature floor to the igm .
our primary goal is to study cosmological shocks which are mostly outside cluster core regions .
we established a temperature floor as a part of our analysis procedure .
the conclusions drawn in this study , hence , should not be significantly weakened by the exclusion of those additional physical processes . while shocks are automatically detected during the simulations by the riemann solver within the hydrodynamics routine ,
there are additional steps necessary to identify and characterize shocks for analysis .
we have done this as a post - processing step using the simulation data at selected epochs .
ideally , explicitly three - dimensional flow motions should be considered in identifying shocks in simulation data . however , to simplify the analysis we used a one - dimensional procedure applied successively in all three primary directions .
a zone was tagged as a `` shock zone '' currently experiencing shock dissipation whenever these three criteria are met : 1 ) @xmath33 , the gradients of gas temperature and entropy have the same sign , 2 ) @xmath34 the local flow is converging ( where @xmath35 is the divergence of three - dimensional velocity field ) , 3 ) @xmath36 , where in each case we define central differences according to the scheme , @xmath37 .
the third condition corresponds to the temperature jump of a mach 1.3 shock .
typically , a shock is represented by a jump spread over @xmath38 tagged zones .
hence , we identified a `` shock center '' within the numerical shock where @xmath35 is minimum and labeled this center as part of a shock surface .
the mach number of shock centers , @xmath39 , was calculated from the temperature jump across shocks , which is given by @xmath40 where @xmath41 and @xmath42 are the postshock and preshock temperatures .
shock centers identified in multiple directional passes were labeled by the maximum @xmath39 .
we followed only those portions of shock surfaces with @xmath2 to avoid confusion from complex flow patterns and shock surface topologies associated with very weak shocks . in the actual simulations ,
the minimum gas temperature was set as the temperature of cmbr , that is , @xmath43 , since photoionization and heating as well as radiative cooling were ignored .
however , considering that significant reionization would have taken place by @xmath44 ( see , e.g. , * ? ? ?
* ) , in post - processing we set the minimum gas temperature at @xmath45 k. the shock properties described below include that consistency adjustment .
figure 1 shows a two - dimensional slice from the @xmath25 simulation isolating a typical group with x - ray emission weighted temperature , @xmath46 kev , and x - ray luminosity , @xmath47 erg s@xmath48 .
the figure shows the locations of identified shocks along with the x - ray emissivity , gas temperature , density , and velocity field distributions .
although the x - ray emissivity distribution looks relatively smooth and round , there are complex accreting flows around the group including three sheets ( dotted line contours in the temperature contour map ) and one filament ( thin solid line contours ) .
a complex topology of shock surfaces surrounding the group can be seen in the temperature contour map and through the locations of identified shocks .
the group as well as the associated sheets and filament are bounded by shocks , but there are several additional shock structures within that represent a variety of converging flow patterns . to illustrate the limitations of conventional spherical accretion concepts , we note that this group provides an example of structures where shocks can form in the core by organized infall flows accreting from filaments and sheets that penetrate deep into the center of potential wells .
the mach number of the surface along the portion of the accretion shock centered at @xmath49 mpc ranges over @xmath50 with a mean value of 3.2 , while that along the portion centered at @xmath51 mpc ranges @xmath52 with a mean value of 6.5 .
although quite strong shocks of mach number of a few , including these , were often found within @xmath53 mpc from the center of clusters and groups , most shocks identified in our simulations are located outside the cores of clusters and groups ( see the next section and figure 5 ) . in order to illustrate how our shock identification scheme works , we plot in figure 2 the flow structure along the horizontal path drawn just below @xmath54 mpc in the shock location plot of figure 1 . as indicated in the upper right panel of the figure , two outer shocks with high mach numbers and three inner shocks with lower mach numbers
were identified along the path .
the existence of these five shocks is also clearly evident in the velocity field and temperature contours in figure 1 .
extensive tests showed that our scheme identifies shock surfaces reliably with a typical error of a few percent in mach number .
in previous studies by @xcite and by other authors , cosmological shock waves were often organized into three categories ; accretion shocks , merger shocks , and internal flow shocks . however , based on close examination of shock locations and the properties of the shocks and their associated flows in our simulations , we suggest instead that it is very informative to classify cosmological shocks into two broad populations that can be conveniently labeled as _ external _ and _ internal _ shocks , depending on whether or not the associated preshock gas was previously shocked ( that is , whether @xmath55 or @xmath56 for the preshock temperature in practice ) .
this binary classification facilitates the understanding of their role in energy dissipation ( see the next section ) . _
external shocks _ surround sheets , filaments and knots , forming when never - shocked , low density , void gas accretes onto those nonlinear structures .
subsequent , _ internal shocks _ are distributed within the regions bounded by external shocks . they are produced by flow motions accompanying hierarchical structure formation inside the bounding shocks . for more refined questions internal shocks can be further divided into three types : 1 ) accretion shocks produced by infall from sheets to filaments and knots and from filaments to knots , 2 ) merger shocks formed during subclump mergers , and 3 ) flow shocks induced by chaotic supersonic motions inside the nonlinear structures .
figure 3 represents a two dimensional slice of a @xmath57 mpc)@xmath58 volume extracted from the @xmath25 grid zone simulation .
it shows the distributions of external and internal shocks in a cluster with x - ray emission - weighted temperature , @xmath59 kev , and x - ray luminosity , @xmath60 erg s@xmath48 , along with the gas density distribution and the velocity field of the inner parts of the slice .
external shocks here define an entire `` cluster complex '' which has dimensions of about @xmath61 .
numerous internal shocks were identified inside the complex .
the two - dimensional velocity field demonstrates that infalls from several associated filaments and sheets form accretion shocks and also induce chaotic flow motions in the medium around the cluster .
figure 4 shows a three - dimensional view of shock surfaces around the same cluster complex .
surfaces of external shocks with high mach numbers , represented with yellow , encompass the complex and filaments associated with it .
several sheets with lower mach numbers , which intersect at the complex , are also clearly visible .
these figures show clearly that the canonical spherical external accretion shock model around a cluster would be far too simple to apply to this sample cluster .
note , in particular , that a `` spherical '' cluster of @xmath62 kev has the first caustic at @xmath63 mpc from the center in our @xmath0cdm model @xcite , around which the external shock would be located .
that is well inside the external , accretion shock in this simulated cluster complex .
we will now estimate some quantitative measures of shock frequency . to start we computed the surface area of identified shocks per logarithmic mach number interval , @xmath64 , normalized by the volume of the simulation box .
this provides an effective inverse comoving mean distance between shock surfaces . in this accounting
each shock center contributes a surface @xmath65 , which is the mean projected area within a three - dimensional zone for random shock normal orientations .
the upper two panels of figure 5 show @xmath64 for external and internal shocks at several epochs in the simulation with @xmath25 grid zones .
table 1 lists the integrated mean shock separation , @xmath66 , along with the mean values of the shock mach number , the shock speed , the preshock sound speed and the preshock gas density .
several important points are apparent .
1 ) the two populations of shocks have distinctive distributions of shock surfaces , @xmath64 , implying they are induced by flows of different characteristics .
the area distribution of external shock surfaces peaks at @xmath67 , extending up to @xmath1 or higher .
in contrast , the comoving area of internal shocks increases to the weakest shocks we identified in our analysis ( @xmath68 ) .
2 ) while the comoving area of external shock surfaces has not changed much since @xmath69 , that of internal shocks has increased significantly .
the former behavior reflects the fact that the `` bounding '' shapes associated with external shocks have evolved towards simpler shapes to almost balance the increasing comoving volumes enclosed . on the other hand both
the volume enclosed and the complexity of internal shocks has increased , so that these shocks now include more area . at the present epoch
the total area of external shock surfaces is @xmath70 times of that of identified internal shocks .
3 ) although the mean mach number is higher for external shocks , the mean shock speed is actually larger for internal shock . in addition , the preshock gas density is significantly higher for internal shocks .
these factors enhance their dynamical importance ( see the next section ) .
4 ) we found that most clusters and groups with @xmath71 kev have shocks within @xmath72 mpc from the centers at present .
the area distribution of these `` cluster shocks '' , shown in the upper right panel of figure 5 , fits best to @xmath73 with @xmath74 in the range of @xmath75 .
their mean mach number is @xmath76 .
the cluster shocks , however , actually account for only a very small fraction of identified shock surfaces .
we emphasize that the statistics for the cluster shocks would have been affected by the finite resolution , @xmath30 kpc , as well as by the exclusion of physical processes like radiative cooling and feedback from galaxies and stars that influence conditions inside cluster cores .
still , it is significant that , compared with the distribution of binary merger shocks studied in @xcite , shocks with higher mach numbers are more common in the environments of clusters and groups in our simulation .
this difference is because our cluster shock population includes accretion shocks created by infall from filaments and sheets to knots and flow shocks generated by chaotic supersonic motions , as well as those induced by hierarchical merging .
especially , internal accretion shocks are strong with a typical mach number of several ( see 2.2 ) .
the lower two panels of figure 5 show @xmath77 for external and internal shocks at @xmath78 in the simulations with different numerical resolutions of @xmath79 grid zones .
table 2 lists the integrated surface area , @xmath80 , along with the mass of the gas that went through shocks at least once , @xmath81 , and the total gas thermal energy inside the computational box , @xmath82 , at the present epoch in the simulations with @xmath83 grid zones .
the shocked gas mass , @xmath84 was estimated by summing the mass of gas with @xmath85 k. we expect that finite resolution would affect the statistics of shocks in two different ways : 1 ) some shocks , especially weak ones with low mach numbers , may not have been `` captured '' in the hydrodynamic part of simulations , and 2 ) they may not have been `` identified '' by our post - processing shock identification scheme , especially in the regions with complex three - dimensional flow structures .
the former causes quantities like @xmath84 and @xmath86 to be underestimated . on the other hand ,
the latter reduces the estimated value of @xmath87 and so the estimation of shock dissipation ( see the next section ) .
we see that @xmath84 and @xmath86 converge rather quickly and their converged values should be within @xmath88 or so of those from the simulation with @xmath25 .
this indicates that the ability of our code to capture shocks is not the major limitation .
the convergence of the mean shock separation , @xmath89 and @xmath90 , is not as obvious from the table .
the difficulties in reaching final convergence in these quantities is evident in figure 5 .
detection of strong shocks with @xmath91 , especially strong external shocks , is relatively robust in the simulation data with @xmath26 or more grid zones .
however , the area spanned by weak shocks continues to increase with resolution , reflecting the increased flow complexity captured inside nonlinear structures .
the quantities of @xmath89 and @xmath90 are , however , also plotted in figure 8 , where it is evident that convergence is underway between the two highest resolution simulations .
it is likely that @xmath92 mpc , and @xmath93 .
it is , then , interesting to note that since the volume of the nonlinear structures where internal shocks are found is @xmath94 of the entire computed volume , the mean distance between internal shock surfaces is @xmath95 mpc within the nonlinear structures , which is comparable to the scale of the nonlinear structure involved .
this conclusion , of course , includes the caveats that only shocks with @xmath96 have been counted and that we have omitted from the simulations physics likely to influence shock formation inside cluster cores . accepting the above limitations , we argue in the next section from estimations of the gas thermal energy dissipated at shocks and the mass passed through external shocks that the identification of dynamically important shocks should be complete within an error of order @xmath97 in the simulation data with @xmath25 grid zones .
as the first step to quantitative estimates of the gas thermal energy and cr energy from dissipation at cosmological shock waves , we defined the following fluxes of mass and energies at each `` shock center '' : 1 ) the gas mass flux incident on shocks , @xmath98 ( @xmath99 ) ; 2 ) the incident kinetic energy flux , @xmath100 ( @xmath101 ) ; 3 ) the thermal energy flux generated at shocks , @xmath102 ( defined below ) ; and 4 ) the cr energy extracted , `` nonthermal dissipation '' , at shocks , @xmath103 ( defined below ) .
here subscripts 1 and 2 stand for preshock and postshock conditions , respectively .
we define thermal energy flux generated at shocks as @xmath104u_2.\ ] ] we point out that the second term inside the brackets subtracts the effect of adiabatic compression occurred at a shock , not simply the thermal energy flux entering the shock ; namely , @xmath105 .
the ratio @xmath106 then defines the efficiency of shock thermalization , which is a function of mach number only , and can be determined from the rankine - hugoniot jump conditions .
the left panel of figure 6 shows @xmath107 .
as expected , @xmath107 increases with mach number , asymptotically approaching @xmath108 for @xmath109 .
we express the cr energy extraction rate as @xmath110 , where @xmath111 measures the efficiency of diffusive shock acceleration for a given mach number . to estimate this efficiency
we have used results of numerical simulations of the nonlinear evolution of cr modified quasi - parallel plane shocks ( e.g. , * ? ? ?
* ; * ? ? ?
the simulations utilize a plasma - physics - based model for the leakage of thermal ions into the cr population , which depends on a single parameter measuring the enhancement of nonlinear scattering waves across the shock transition . that parameter is reasonably well limited by the theory and by plasma simulations ( see , e.g. , * ? ? ?
* ) , and , in the dsa simulation we used , typically limits the fraction of protons injected into the cr population at shocks to be of order @xmath112 .
the cr acceleration efficiency characteristics of our dsa model are broadly consistent with other widely used theoretical and numerical studies of nonlinear cr shocks ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in order to apply properties of our dsa model to cosmological shock waves , we calculated proton acceleration and accompanying cr - modified flow evolution for shocks with @xmath113 km s@xmath48 propagating into media of @xmath114 k , assuming bohm - type diffusion for the crs ( for details see * ? ? ? * ) .
as the shock structures evolved we determined the ratio of the total cr energy extraction during the evolution of a shock to the kinetic energy passed through the shock according to @xmath115}.\ ] ] the values of @xmath116 quickly reached approximate `` time - asymptotic '' values , after which the shock structures evolved approximately `` self - similarly . '' these asymptotic values of @xmath116 were taken as our estimates for @xmath117 .
the left panel of figure 6 shows the resulting cr acceleration efficiency , @xmath117 , along with the gas thermalization efficiency @xmath107 .
both @xmath107 and @xmath117 increase with mach number . in
strong , high mach number shocks , @xmath117 approaches the asymptotic value of @xmath118 .
we see that @xmath107 is somewhat larger than @xmath117 over all mach numbers . on the other hand
, we note that @xmath119 was computed for purely gasdynamical shocks ; that is , without accounting for energy removal into crs and enhanced adiabatic compression within the cr shock precursor .
if nonlinear dynamical feedback of crs were included self - consistently in estimating @xmath107 , the resulting @xmath107 would be somewhat smaller than that shown in figure 6 . from the above gas mass and kinetic energy fluxes at each shock center , we calculated the associated fluxes through surfaces of shocks with mach number between @xmath120 and @xmath121 at different redshifts ; @xmath122 and @xmath123 , respectively .
we also calculated the similarly defined fluxes of gas thermal and cr energies dissipated at shocks as @xmath124 and @xmath125 , respectively .
the right panel of figure 6 illustrates @xmath126 per unit comoving volume for external and internal shocks at different redshifts in the simulation with @xmath25 grid zones .
a noticeable point is that the kinetic energy flux , and also the gas mass flux through external shocks were larger in the past .
this is because the preshock gas density was larger in the past .
however , the kinetic energy flux through internal shocks has been more or less constant since @xmath127 , and was smaller before that . to provide measures of the roles of shocks over time we integrated from @xmath128 to @xmath78 the gas mass and kinetic energy that passed through shock surfaces and the gas thermal and cr energies dissipated at shock surfaces as @xmath129 where the subscript @xmath130 stands for the four fluxes defined above .
the quantities were normalized either to the shocked gas mass , @xmath131 , for mass or to the total gas thermal energy inside the computational box , @xmath132 , for energies ( see table 2 ) .
we also summed these time - integrated measures to calculate the associated global shock - processed quantities , @xmath133 d \log m.\ ] ] @xmath134 and @xmath135 not only measure the total mass that passed through shocks in the @xmath128 to @xmath78 interval , but also , by way of their normalization , measure the mean number of times that gas has been subjected to shock dissipation .
meanwhile , @xmath136 and its derivative measure the total kinetic energy that has been subject to shock dissipation .
on the other hand , @xmath137 and @xmath138 compare the total thermal and cr energies that have resulted from shock dissipation .
figure 7 shows @xmath139 and @xmath140 for external and internal shocks in the simulation with @xmath25 grid zones .
again , several important points are apparent .
1 ) the plots for @xmath135 and @xmath141 indicate that more mass has passed through internal shocks than external shocks .
with @xmath142 for internal shocks , the mass inside the nonlinear structures of sheets , filaments and knots has been shocked , on average , twice or so by internal shocks from @xmath128 to 0 .
but we note that @xmath135 for internal shocks increases to @xmath143 , the lowest mach number we kept , and weak shock identification is not fully converged yet with this numerical resolution .
so the above value of 2.2 should be regarded as the lower limit .
2 ) on the other hand , since gas enters the nonlinear structures by passing through external shocks , @xmath144 for external shocks ( the mass passed through external shocks from @xmath128 to 0 ) should match the increase in the shocked gas mass , @xmath84 , from @xmath128 to 0 . in the simulation with @xmath25 grid zones , we get @xmath145 for external shocks , while @xmath146 is expected with @xmath147 and @xmath148 ( where @xmath149 is the total gas mass in the computational box ) .
although a fraction of the hot gas may have been heated above @xmath150 k by adiabatic compression in void regions or by weak shocks with @xmath151 , some of this discrepancy could be due to the under - counting of external shocks .
so we estimate an error of @xmath88 or so in the completeness of the identification of external shocks .
3 ) the lower two panels of figure 7 show that internal shocks play a more important role than external shocks in the dissipating energy associated with structure formation .
specifically , internal shocks with @xmath152 account for @xmath7 of dissipation . while the thermal energy generation peaks for shocks in the range @xmath153 , the cr energy extraction peaks in the range @xmath154 .
@xcite identified shocks with @xmath155 as the most important in gas thermalization .
our result indicates a somewhat lower mach number range , because we found more weaker shocks in our simulation data .
this difference is probably due to our greater resolution ( see , e.g. , figure 5 ) , as well as due to improved shock capturing in the present code and a more sophisticated shock identification scheme .
4 ) the amount of thermal energy generated at shocks from @xmath128 to 0 was computed to be @xmath156 times the current gas thermal energy in the simulation with @xmath25 grid zones . since
no other heating or cooling physics was included , there are two likely contributors to the difference .
first , we attribute some of the difference to adiabatic expansion of nonlinear structures after formation .
expansion of @xmath157 along each dimension would be enough to reduce the gas thermal energy by a factor of 1.5 .
in addition , some of the discrepancy could be the result of multiple counting of internal shocks in the regions with complex flow structures .
so again , we put an error of a few @xmath97 , at most , in the completeness of the identification of shocks .
5 ) with the dsa model we adopted @xcite the ratio of the cr to gas thermal energies dissipated at cosmological shock waves with the mach number greater than 1.5 is @xmath158 .
6 ) there are many shocks with @xmath159 , but they are not important in energetics . hence , although we ignored the shocks with @xmath151 in this work , that should not significantly impact quantitative energetics results presented here .
figure 8 shows @xmath160 through all shock surfaces in the simulations with different resolution of @xmath83 grid zones .
note that @xmath161 s were normalized with @xmath84 and @xmath162 for a given resolution ( table 2 ) . yet , for instance , @xmath136 , the integrated kinetic energy flux through shock surfaces , decreases in the plot by @xmath163 from @xmath25 to @xmath14 and by @xmath164 from @xmath14 to @xmath26 . on the other hand ,
@xmath162 , the total gas thermal energy inside the computational box , decreases only by @xmath165 from @xmath25 to @xmath14 and @xmath166 from @xmath14 to @xmath26 .
this reinforces the statements in the previous section that the error in quantitative assessments of this paper comes mostly from shock identification in the post - processing analysis rather than shock capturing in the code .
this also shows that our estimation of an error of order @xmath97 in the data of the @xmath25 simulation is consistent with resolution convergence .
we identified and studied shock waves with mach number @xmath2 in a set of cosmological n - body / hydrodynamic simulations for a @xmath0cdm universe in a cubic box of comoving size @xmath24 mpc . to facilitate the analysis of their properties , the cosmological shock waves were classified as external and internal shocks , depending on whether or not the preshock gas was previously shocked .
external shocks form around outermost surfaces that encompass nonlinear structures , so they are by nature accretion shocks that decelerate the previously unshocked intergalactic gas infalling toward sheets , filaments and knots .
internal shocks are produced within those nonlinear structures by accretion flows of previously shocked gas from sheets to filaments and knots and from filaments to knots , by merging of subclumps , or by chaotic flow motions induced in the course of hierarchical clustering .
for all shocks of @xmath2 identified in the simulation of highest resolution , the mean distance between shock surfaces , the inverse indicator of shock occurrence , is @xmath167 mpc at present .
further , external shocks are more extensive , with their surface area @xmath70 times larger than that of identified internal shocks at present . with the volume of nonlinear structures which is @xmath94 of the total volume ,
the mean distance between internal shock surfaces is @xmath95 mpc within nonlinear structures at present . although external shocks typically have higher mach numbers , internal shocks have higher shock speed and higher preshock gas density . as a result , internal shocks are responsible for @xmath168 of gas thermalization and for @xmath169 of cr acceleration at shocks , and they process about 6 times more gas mass through shock surfaces than external shocks do from @xmath128 to @xmath78 . internal shocks with @xmath170
are especially important in energy dissipation , contributing @xmath7 of the total . by adopting a model of cr proton acceleration based on nonlinear diffusive shock simulations @xcite ,
our study predicts that the ratio of the cr to gas thermal energies dissipated at all cosmological shocks through the history of the universe could be @xmath8 .
due to long cr proton trapping times and energy loss lifetimes , they should fill the volumes inside filaments and sheets as well as in clusters and groups .
short electron lifetimes , however , lead that population to depend on other factors @xcite . the existence of substantial cr populations could have affected the evolution and the dynamical status of the large scale structure of the universe ( see , e.g. , * ? ? ?
the examination of results from simulations of different resolutions showed that the convergence with resolution in shock identification is slower than in shock capture within the simulation itself .
consistency checks and resolution convergence analysis lead to error estimates of order of @xmath97 in our quantitative estimates of the accounting of and energy dissipation in cosmological shock waves .
strong , external shocks are energetically less important than moderate strength internal shocks .
however , it is important to keep in mind that the large curvature radii and long life times of these shocks make them viable candidates to accelerate crs to ultra high energies of several @xmath171 ev @xcite .
the mach number distribution and the amount of energy dissipation at cosmological shocks have significant implications for several cosmological observations such as radio and @xmath9-ray emissions as well as x - ray emission from the icm and contribution to the cosmic @xmath9-ray background from crs accelerated at these shocks ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these important issues , along with the observational manifestation of cosmological shock waves , will be considered in future studies .
dr and hk were supported by kosef through astrophysical research center for the structure and evolution of cosmos ( arcsec ) . eh and twj were supported by the nsf ( ast00 - 71167 ) , nasa ( nag5 - 10774 ) and the university of minnesota supercomputing institute .
numerical simulations utilized `` the grand challenge program '' of the kisti supercomputing center .
we thank peter l. biermann and an anonymous referee for constructive comments on the manuscript .
lcccccccccc 0 & 4.4 & 2.1 & 8.0 & 3.2 & 123 & 226 & 15.3 & 82 & 1.05 & 6.78 + 0.2 & 4.4 & 2.3 & 8.1 & 3.3 & 123 & 230 & 15.3 & 83 & 1.12 & 7.15 + 0.5 & 4.5 & 2.8 & 8.0 & 3.3 & 122 & 231 & 15.3 & 83 & 1.25 & 7.86 + 1 & 5.0 & 3.7 & 7.5 & 3.4 & 114 & 214 & 15.3 & 76 & 1.48 & 8.87 + 1.5 & 5.7 & 5.0 & 7.0 & 3.4 & 107 & 196 & 15.3 & 69 & 1.79 & 10.3 + 2 & 6.8 & 6.6 & 6.5 & 3.4 & 100 & 177 & 15.3 & 62 & 2.14 & 10.9 + | we study the properties of cosmological shock waves identified in high - resolution , n - body / hydrodynamic simulations of a @xmath0cdm universe and their role on thermalization of gas and acceleration of nonthermal , cosmic ray ( cr ) particles .
external shocks form around sheets , filaments and knots of mass distribution when the gas in void regions accretes onto them . within those nonlinear structures , internal shocks
are produced by infall of previously shocked gas to filaments and knots , and during subclump mergers , as well as by chaotic flow motions .
due to the low temperature of the accreting gas , the mach number of external shocks is high , extending up to @xmath1 or higher .
in contrast , internal shocks have mostly low mach numbers .
for all shocks of @xmath2 the mean distance between shock surfaces over the entire computed volume is @xmath3 mpc at present , or @xmath4 mpc for internal shocks within nonlinear structures .
identified external shocks are more extensive , with their surface area @xmath5 times larger than that of identified internal shocks at present . however , especially because of higher preshock densities , but also due to higher shock speeds , internal shocks dissipate more energy .
hence , the internal shocks are mainly responsible for gas thermalization as well as cr acceleration .
in fact , internal shocks with @xmath6 contribute @xmath7 of the total dissipation . using a nonlinear diffusive shock acceleration model for cr protons
, we estimate the ratio of cr energy to gas thermal energy dissipated at cosmological shock waves to be @xmath8 through the history of the universe .
our result supports scenarios in which the intracluster medium contains energetically significant populations of crs . |
in this paper we present new evidence bearing on the formation of disks and halos of spiral galaxies through the accretion of small companions .
this process is generally indicated as the nurture of galaxies .
evidence in support of it comes from various directions .
the asymmetric shape of stellar disks ( zaritsky 1995 and zaritsky & rix 1997 ) and the morphological and kinematic lopsidedness observed in the h0.1emi density distributions and velocity fields of spirals ( verheijen 1997 , swaters et al . 1999 ) may have originated from recent minor mergers .
furthermore , there is an increasingly large number of galaxies which in h0.1emi show either peculiar features or clear signs of interactions with small companions ( sancisi 1999a and b ) .
this indicates that galaxies often are in an environment where material for accretion is available .
cold extra - planar gas has been found in several spiral galaxies .
the best examples are those of ngc 891 , ngc 2403 and ugc 7321 ( cf .
oosterloo et al . ,
fraternali et al . ,
matthews , this volume ) .
the origin of this gas is not known .
it has been suggested that it may , at least partly , be the product of galactic fountains . but some of its structural properties suggest that it may have originated from minor mergers .
recently , clear evidence that accretion events play an important role has come from studies of the distribution and kinematics of stars in the milky way halo .
the discovery of the sagittarius dwarf galaxy ( ibata et al .
1994 ) is regarded as proof that accretion is still taking place at the present time . since
such minor merger remnants retain information about their origin for a long time ( helmi & white 2000 ) studies of the distribution and kinematics of `` stellar streams '' can in principle be used to trace the merger history of the milky way ( helmi & de zeeuw 2001 ) .
such `` stellar streams '' are not only seen in the milky way , but have also been discovered in the local group galaxy m 31 ( ibata et al .
2001 , ferguson et al .
2002 , mcconnachie et al .
the substructure in the halo of m 31 is another piece of clear evidence that minor mergers still take place .
such events are difficult to trace in more distant galaxies , where we can not observe individual stars .
other means are needed for detecting the signature of accretion . in this respect ,
the use of h0.1emi is very powerful as it can image interactions very effectively by studying the h0.1emi distributions and kinematics .
the latter is particularly useful for modelling .
examples can be found in sancisi ( 1999a ) .
the improved sensitivity of modern synthesis radio telescopes brings within reach the detection of faint h0.1emi signatures of accretion events and we expect that new observations of nearby and also more distant galaxies will reveal these in the coming decade . to further illustrate this point we here present a few examples of such signatures : ngc 3359 , ngc 4565 and ngc 6946 .
all three galaxies have been observed recently with the westerbork synthesis radio telescope ( wsrt ) using the new front - end and correlator providing a much improved sensitivity .
we will discuss each case individually below .
details of the observations and some of these results have already been reported by van der hulst & sancisi ( 2004 ) .
ngc 3359 is a nearby barred spiral galaxy ( hubble type sbc ) which has been observed in h0.1emi by broeils ( 1992 ) as part of a study of the mass distribution of a sample of nearby spiral galaxies .
it has a total mass of @xmath0 m@xmath1 and an h0.1emi mass of @xmath2 m@xmath1 ( broeils & rhee , 1997 , adjusted for a hubble constant of 72 km / s / mpc ) .
it has well developed spiral structure both in the optical and in h0.1emi .
kamphuis & sancisi ( 1994 , see also sancisi 1999a ) pointed out the presence of an h0.1emi companion which appears distorted and has a long tail which may connect to the h0.1emi disk of ngc 3359 .
this observation already indicated the possible accretion of gas by a large galaxy .
our new , more sensitive observations ( rms noise of 0.85 mjy / beam for velocity and spatial resolutions of 10 km s@xmath3 and 30@xmath4 respectively ) are shown in figure 1 ( left panel ) and convincingly display an h0.1emi connection between the distorted h0.1emi companion and the main galaxy .
the mass of the h0.1emi companion is @xmath5 m@xmath1 or 2.4% of the h0.1emi mass of ngc 3359 .
the h0.1emi distribution of the companion is clearly distorted and shows a tail pointing towards and connecting with the outer spiral structure of ngc 3359 .
no clear optical counterpart of the h0.1emi companion has yet been found .
the velocity structure of the h0.1emi companion and the connecting h0.1emi fits in very well with the regular velocity field of ngc 3359 .
this is shown in the right panel of figure 1 where we display the emission in the individual channels superposed on the total h0.1emi image of ngc 3359 .
contours of different shades of grey ( low velocities are dark , high velocities are light ) denote the outer edge of the h0.1emi emission in each of the velocity channels and thus display the basic kinematics of the h0.1emi without any further analysis of individual velocity profiles .
the regularity of the velocities suggests that the process has been going on slowly for at least one rotational period which is of the order of 1.7 gy .
ngc 4565 is a large edge - on galaxy of hubble type sb which was first observed in h0.1emi by sancisi ( 1976 ) in an early search for galaxies with warped h0.1emi disks .
rupen ( 1991 ) observed ngc 4565 with much higher resolution and presented a detailed study of the kinematics and the warp .
ngc 4565 has a small optical companion 6@xmath6 to the north of the center , f378 - 0021557 , which has @xmath7 m@xmath1 of h0.1emi compared to an h0.1emi mass of @xmath8 m@xmath1 for ngc 4565 ( using a distance of 17 mpc ) .
an h0.1emi detection of this companion called ngc 4565a has also been reported by rupen ( 1991 ) .
another h0.1emi companion , ngc 4562 , somewhat larger in h0.1emi ( @xmath9 m@xmath1 ) and brighter optically can be found 15@xmath6 to the south - west of the center of ngc 4565 .
the h0.1emi distribution , derived from a new sensitive wsrt observation by dahlem ( priv . comm . ) , is shown in figure 2 superposed on the dss .
the asymmetric warp is clearly visible .
inspection of individual channel maps brings to light that in addition to the warp the h0.1emi distribution shows additional , low surface brightness emission to the north of the center , in the direction of the faint companion f378 - 0021557 . the h0.1emi emission in the velocity range from 1250 to 1290 km
s@xmath3 ( close to the velocity of f378 - 0021557 and to the systemic velocity , 1230 km s@xmath3 , of ngc 4565 ) clearly shows distortions above the plane pointing towards the companion .
this is best seen in figure 3 which shows four channel maps chosen at velocities in this range . in these maps
one can clearly see the h0.1emi layer bending towards f378 - 0021557 , indicating a connection between f378 - 0021557 and a strong disturbance in the h0.1emi disk of ngc 4565 .
this disturbance is not associated with the warp . however
, this bending of the h0.1emi layer , undoubtedly caused by the companion , is remarkably similar tot the outer h0.1emi warping . one wonders whether the type of interaction we are witnessing here is the mechanism also responsible for the creation of a warp .
this interaction between ngc 4565 and its companion will eventually lead to a merger of the latter with ngc 4565 . to further elucidate the connection between f378 - 0021557 and ngc
4565 we show a position - velocity diagram along the h0.1emi connection in figure 4 .
this position - velocity cut has been taken parallel to the minor axis of ngc 4565 through f378 - 0021557 and clearly shows the velocity continuity of the h0.1emi disturbance in the disk of ngc 4565 and f378 - 0021557 .
ngc 6946 is a bright , nearby spiral galaxy of hubble type scd which has been studied in h0.1emi numerous times ( rogstad et al .
1973 , tacconi & young 1986 , kamphuis 1993 ) .
it was in this galaxy that kamphuis and sancisi ( 1993 ) found the first evidence for an anomalous velocity h0.1emi component which they associated with outflow of gas from the disk into the halo as a result of stellar winds and supernova explosions .
evidence for such a component is now being found in more galaxies as discussed by fraternali et al .
( 2001 , 2002 , and also this volume ) .
a detailed study of the anomalous h0.1emi and the structure in the h0.1emi disk is being carried out by boomsma et al .
( this volume ) on the basis of very sensitive observations with the wsrt ( rms of 0.5 mjy / beam for spatial and velocity resolutions of 60@xmath4 and 5 km s@xmath3 respectively ) .
here we use a low resolution ( 60@xmath4 ) version of these data .
figure 6 shows a total h0.1emi image of ngc 6946 down to column density levels of @xmath10 @xmath11 . to the west two small companion galaxies
can be seen .
the most intruiging feature is the faint whisp to the north - west of the h0.1emi disk of ngc 6946 .
this faint h0.1emi extension can only be brought out at this resolution and appears to form a faint h0.1emi filament which blends smoothly ( also kinematically ) with the h0.1emi disk of ngc 6946 at a position some 11@xmath6 ( or 19 kpc ) south of the tip of the filament .
there is no detected connection with the two companion galaxies farther to the west .
the spatial and velocity structure of the object are so regular , yet only connected to the main h0.1emi disk at one side that we prefer an explanation in terms of a tidally stretched , infalling h0.1emi object .
so this looks like yet another example of accretion of small amounts of gas onto a large h0.1emi disk .
similar examples are perhaps the filament discovered in ngc 2403 ( fraternali et al . 2001 , 2002 and also this volume ) , a long h0.1emi filament in m 33 ( van der hulst , unpublished ) and the extra - planar filaments in the northern part of the h0.1emi halo of ngc 891 ( fraternali et al .
, this volume ) .
we have shown three cases with strong evidence for the accretion of small amounts of h0.1emi .
these are certainly not unique .
there are several more cases known ( sancisi 1999a and b ) . furthermore , there are cases with such faint features that can only be seen in sensitive h0.1emi observations as the h0.1emi masses involved are rather modest .
we therefore expect that with the increased sensitivity of modern synthesis radio telescopes , more examples will be discovered in the coming decade .
there probably is a range of h0.1emi masses for these accretion events as is already apparent from the six cases mentioned here : ngc 891 , ngc 2403 , ngc 3359 , ngc 4565 , ngc 6946 and m 33 . what is the effect of accretion on the main galaxy
this can influence local star formation in the disks and starbursts .
for instance , there may very well be a connection with the star formation activity in the disks of galaxies such as ngc 6946 and ngc 2403 .
the infall of gas may be at the origin of the extra - planar gas and gaseous halos recently discovered in spiral galaxies .
also , it may affect the structure of galactic disks in the outer parts and possibly also contribute to the formation of the outer layers and of warps in particular .
it is quite clear that future sensitive and detailed studies of the h0.1emi in nearby galaxies will provide a more complete census of the phenomena discussed in this paper and enable us to address these issues further and obtain more definitive answers .
in particular , it should be possible to obtain estimates of the mean gas accretion rate in galaxies .
broeils , a. h. , 1992 , phd thesis , university of groningen broeils , a. h. , rhee , m. -h . , 1997 , , 324 , 877 ferguson , a. m. n. , irwin , m. j. , ibata , r. a. , lewis , g. f. , & tanvir , n. r. 2002 , , 124 , 1452 fraternali , f. , oosterloo , t. , sancisi , r. , & van moorsel , g. 2001 , , 562 , l47 fraternali , f. , van moorsel , g. , sancisi , r. , & oosterloo , t. 2002 , , 123 , 3124 helmi , a. & tim de zeeuw , p. 2000 , , 319 , 657 helmi , a. & white , s. d. m. 2001 , , 323 , 529 ibata , r. , irwin , m. , lewis , g. , ferguson , a. m. n. , & tanvir , n. 2001 , nature , 412 , 49 ibata , r. a. , gilmore , g. , & irwin , m. j. 1994 , nature , 370 , 194 kamphuis , j. 1993 , phd thesis , university of groningen kamphuis , j. & sancisi , r. 1993 , , 273 , l31 kamphuis , j. & sancisi , r. 1994 , panchromatic view of galaxies .
their evolutionary puzzle , eds . g. hensler , c. theis and j.s .
gallagher , editions frontiers , p. 317 mcconnachie , a. w. , irwin , m. j. , ibata , r. a. , ferguson , a. m. n. , lewis , g. f. , & tanvir , n. 2003 , , 343 , 1335 rogstad , d. h. , shostak , g. s. , & rots , a. h. 1973 , , 22 , 111 rupen , m. p. 1991
, , 102 , 48 sancisi , r. 1976 , , 53 , 159 sancisi , r. 1999a , iau symp . 186 : galaxy interactions at low and high redshift , eds .
barnes and d. b. sanders , p. 71
sancisi , r. 1999b , , 269 , 59 tacconi , l. j. & young , j. s. 1986 , , 308 , 600 van der hulst , j. m. & sancisi , s. 2004 , iau symp . 217 : recycling intergalactic and interstellar matter , eds .
duc , j. braine and e. brinks , p. 122 zaritsky , d. 1995 , , 448 , l17 zaritsky , d. & rix , h. 1997 , , 477 , 118 | evidence for the accretion of material in spiral galaxies has grown over the past years and clear signatures can be found in h0.1emi observations of galaxies .
we describe here new detailed and sensitive h0.1emi synthesis observations of a few nearby galaxies ( ngc 3359 , ngc 4565 and ngc 6946 ) which show that indeed accretion of small amounts of gas is taking place .
these should be regarded as examples illustrating a general phenomenon of gas infall in galaxies .
such accretion may also be at the origin of the gaseous halos which are being found around spirals .
probably it is the same kind of phenomenon of material infall as observed in the stellar streams in the halo and outer parts of our galaxy and m 31 |
we have presented high resolution data for five dla systems analyzing the velocity profiles and the column densities . the systems at @xmath1 = 3.0248 toward q 0347383 and @xmath2 = 2.5226 toward q 1213 + 093 have similar abundance patterns .
especially the nitrogen to silicon relative abundance is smaller than the solar value ( which is true also for the system at @xmath1 = 2.6184 toward q 0913 + 072 ) .
the system at @xmath1 = 2.6184 toward q 0913 + 072 is the most metal - deficient dla system known by now .
we find [ o / h ] @xmath5 @xmath32.7 and [ fe / si ] @xmath32 @xmath30.8
. a better limit on iron is needed .
indeed it would be difficult to reconcile this pattern with standard chemical evolution models .
it is interesting to note that this system has a metal content similar to that of ly@xmath0 clouds with log @xmath15 ( ) @xmath14 15 ( songaila & cowie @xcite ) . from a sample of 26 dla systems gathered from the literature ,
we show that the velocity widths of the @xmath61808 and @xmath61608 absorptions are closely correlated .
this suggests that the physical conditions in the neutral phase are quite homogeneous and that large relative metallicity variations and thus large depletions into dust grains are unlikely .
we discuss the systems with velocity widths larger than 200 km s@xmath8 and conclude that they are most of the time composed of several sub - systems .
we argue , from the distribution of velocity widths and the correlation between the asymmetry parameter and the velocity width , that rotation motions may be present in sub - systems on scales smaller than 150 km s@xmath8 . the velocity width of the strongest part of the absorption decreases with increasing redshift .
this suggests that the neutral regions get denser and exhibit faster motions with time .
boiss p. , le brun v. , bergeron j. , deharveng j.m . , 1998 , a&a 333 , 841 cowie l.l . , songaila a. , 1995 , apj 453 , 596 cowie l.l . , songaila a. , kim t .- s .
, hu e.m . , 1995 , aj 109 , 1522 djorgovski s.g . ,
pahre m.a .
, bechtold j. , elston r. , 1996 , nat 382 , 234 fall s.m . , pei y.c . , 1993 , apj 402 , 479 fontana a. , ballester p. , 1995
, the messenger 80 , 37 franois p. , 1987
, a&a 176 , 294 gratton r.g . , sneden c. , 1991 , a&a 241 , 501 haehnelt m.g . ,
steinmetz m. , rauch m. , 1998 , apj 495 , 647 junkkarinen v. , hewitt a. , burbidge g. , 1991 , apjs 77 , 203 kauffmann g. , 1996 , mnras 281 , 475 kulkarni v.p . , fall s.m .
, truran j.w . ,
1997 , apj 484 , l7 lanzetta k.m . ,
wolfe a.m. , turnshek d.a .
, 1995 , apj 440 , 435 le brun v. , bergeron j. , boiss p. , deharveng j.m . , 1997 ,
a&a 321 , 733 levshakov s.a . ,
varshalovich d.a . , 1985 , mnras 212 , 517 lipman k. , 1995 , phd thesis , cambridge university lu l. , savage b.d . ,
tripp t.m . ,
meyer d.m . , 1995 , apj 447 , 597 lu l. , sargent w.l.w . , womble d.s . , barlow t.a . , 1996a , apj 457 , l1 lu l. , sargent w.l.w . ,
barlow t.a . ,
churchill c.w . , vogt s.s .
, 1996b , apjs 107 , 475 lu l. , sargent w.l.w . ,
barlow t.a . , 1997 ,
apj 484 , 131 lu l. , sargent w.l.w . ,
barlow t.a . , 1998 , aj 115 ,
55 matteucci f. , molaro p. , vladilo g. , 1997 , a&a 321 , 45 meyer d.m . , york d.g . , 1987 ,
apj 319 , l45 meyer d.m .
, welty d.e .
, york d.g .
, 1989 , apj 343 , l37 meyer d.m .
, lanzetta k.m . , wolfe a.m. , 1995 , apj 451 , l13 molaro p. , centurion m. , vladilo g. , 1998 , mnras 293 , l37 morton d.c . , 1991 , apjs 77 , 119 morton d.c . , chen j .- s . ,
wright a.e . ,
peterson b.a . ,
jauncey d.l . , 1980 ,
mnras 193 , 399 mller p. , warren s.j .
, 1993 , a&a 270 , 43 pei y.c .
, fall s.m . , bechtold j. , 1991 , apj 378 , 6 petitjean p. , bergeron j. , puget j.l .
, 1992 , a&a 265 , 375 petitjean p. , rauch m. , carswell r.f .
, 1994 , a&a 291 , 29 pettini m. , smith l.j .
, hunstead r.w . , king d.l . , 1994 , apj 426 , 79 pettini m. , lipman k. , hunstead r.w . , 1995 , apj 451 , 100 pettini m. , king d.l . ,
smith l.j . ,
hunstead r.w . , 1997a , apj 478 , 536 pettini m. , smith l.j .
, king d.l . , hunstead r.w . , 1997b , apj 486 , 665 pierre m. , shaver p.a . , robertson j.g . , 1990 ,
a&a 235 , 15 prochaska j.x . , wolfe a.m. , 1996 , apj 470 , 403 prochaska j.x . , wolfe a.m. , 1997a , apj 474 , 140 prochaska j.x . , wolfe a.m. , 1997b , apj 487 , 73 prochaska j.x .
, wolfe a.m. , 1998 , astro - ph/9805293 sargent w.l.w . ,
steidel c.c . ,
boksenberg a. , 1988 , apjs 68 , 539 savage b.d . ,
sembach , k.r . , 1996 , ara&a 34 , 279 sembach k.r . , savage b.d . , 1996 , apj 457 , 211 songaila a. , cowie l.l . , 1996 , aj 112 , 335 srianand r. , petitjean p. , 1998
, astro - ph/9804036 steidel c.c .
, 1990 , apjs 72 , 1 storrie - lombardi l.j . ,
mcmahon r.g .
, irwin m.j . , 1996 , mnras 283 , l79 turnshek d.a . , wolfe a.m. , lanzetta k.m . , et al . , 1989 ,
apj 344 , 567 turnshek d.a . , 1997 , in structure and evolution of the intergalactic medium from qso absorption line systems , eds .
p. petitjean and s. charlot ( paris : editions frontires ) , 263 viegas s.m . , 1995 ,
mnras 276 , 268 vladilo g. , 1998 , apj 493 , 583 vladilo g. , molaro p. , matteucci f. , centurin m. , 1997 , in the early universe with the vlt , ed . j. bergeron ( berlin : springer ) , 430 warren s.j .
, mller p. , 1996 , a&a 311 , 25 williger g.m .
, carswell r.f .
, webb j.k . , boksenberg a. , smith m.g . , 1989 ,
mnras 237 , 635 wolfe a.m. , 1995 , in qso absorption lines , ed .
g. meylan ( berlin : springer ) , 13 wolfe a.m. , prochaska j.x . , 1998 ,
apj 494 , l15 wolfe a.m. , turnshek d.a .
, smith h.e .
, cohen r.d . , 1986 ,
apjs 61 , 249 wolfe a.m. , fan x .-
m . , tytler d. , et al .
, 1994 , apj 435 , l101 wolfe a.m. , lanzetta k.m .
, foltz c.b .
, chaffee f.h . , 1995 ,
apj 454 , 698 | we report on high spectral resolution observations of five damped lyman-@xmath0 absorbers : @xmath1 = 3.0248 toward q 0347383 , @xmath2 = 2.1411 and 2.8112 toward q 0528 250 , @xmath1 = 2.6184 toward q 0913 + 072 , and @xmath1 = 2.5226 toward q 1213 + 093 .
line velocity profiles and heavy element abundances are discussed .
nitrogen is found to have abundances less than silicon in the systems toward q 0347@xmath3383 , q 0913 + 072 , and q 1213 + 093 .
the absorber toward q 0913 + 072 is the most metal - deficient damped system known , with [ fe / h ] @xmath4 @xmath33.2 .
the simple kinematical structure of the metal absorptions makes this system ideal to discuss the [ o / si ] and [ n / o ] ratios .
we find [ o / h ] @xmath5 @xmath32.7 and @xmath32.7 @xmath4 [ si / h ] @xmath4 @xmath32.2 . by combining these data with information gathered in the literature
, we study the kinematics of the low and high ionization phases in a sample of 26 damped lyman-@xmath0 systems in the redshift range [email protected] .
we note a strong correlation between the velocity broadenings of the @xmath61808 and @xmath61608 lines whatever the line optical depth , implying that the physical conditions are quite homogeneous in the sample .
statistically this shows that large variations of abundance ratios and thus large variations of depletion into dust grains are unlikely .
the velocity broadening of the absorption lines , @xmath7 , is correlated with the asymmetry of the lines for @xmath7 @xmath4 150 km s@xmath8 .
the broader the line the more asymmetric it is , as expected in case rotation dominates the line broadening
. however this correlation does not hold for larger @xmath7 suggesting that evidence for rotational motions is restricted to velocity broadenings @xmath7 @xmath4 150 km s@xmath8 . the systems with @xmath7 @xmath9 200 km s@xmath8 are peculiar with kinematics consistent with random motions .
they show sub - systems as those expected if the objects are in the process of merging .
although the sample is not large enough to draw firm conclusions , there is a trend for the mean velocity broadening to decrease with redshift from 80 km s@xmath8 at @xmath10 @xmath4 2.2 to 50 km s@xmath8 at @xmath10 @xmath9 2.2 .
the kinematics of the low and high ionization species are found to be statistically correlated , though the high - ionization phase has a much more disturbed kinematical field than the low - ionization phase .
this should be taken into account in any model of high redshift damped lyman-@xmath0 systems . [
cols="<,^,^,^,^,^,^,^ " , ] references : ( 1 ) lu et al .
@xcite ; ( 2 ) meyer et al .
@xcite ; ( 3 ) wolfe @xcite ; ( 4 ) prochaska & wolfe @xcite ; ( 5 ) prochaska & wolfe @xcite ; ( 6 ) pettini et al . @xcite ; ( 7 ) wolfe et al .
@xcite ; ( 8) this work ; ( 9 ) lu et al .
@xcite ; ( 10 ) lu et al
. @xcite .
[ kinetab ] the velocity broadening of the @xmath111608,2586 ( at @xmath12 = 0.5 or @xmath13 @xmath14 0.7 ) and @xmath61808 ( at @xmath12 = 0.9 or @xmath13 @xmath14 0.1 ) line profiles are compared in fig . [ patsfss ] .
it can be seen that they correlate very closely .
this implies that we can use the iron velocity interval that is easier to measure because of larger optical depths .
the iron lines are observed in a large number of systems and we are able to compare velocities measured _ with the same ion_. prochaska & wolfe ( @xcite ) made used of several transitions because @xmath61808 line is not detected in all the systems .
since the correlation is very good , it implies that the physical conditions are quite homogeneous in the sample .
first we can be confident that the criteria are well choosen to probe the densest regions where iron and silicon are both predominantly singly ionized . if it were not the case we would expect large variations in the @xmath15()/@xmath15 ( ) ratio as persists much longer than when the column density decreases . indeed the @xmath15()/@xmath15 ( )
ratio can be larger than ten for log @xmath15 ( ) @xmath14 18 @xmath16 ( see petitjean et al .
@xcite ) .
it also suggests that large variations of abundance ratios and/or large depletion into dust grains are unlikely .
these conclusions are consistent with detailed studies of dla systems ( lu et al .
@xcite , prochaska & wolfe @xcite , pettini et al .
@xcite , vladilo 1998 ) . we can infer from this correlation that , statistically
, the local optical depth of @xmath111608,2586 is approximately seven times larger than the local optical depth of @xmath61808 .
hence , using the oscillator strengths given by savage & sembach ( @xcite ) , the column density is on average three times larger than the column density .
the solar metallicities of iron and silicon are similar which implies that in these systems and on average , [ si / fe ] @xmath14 0.5 in the gas phase .
fig .
[ pats_f ] ( upper panels ) shows the histograms of the velocity broadening measured from at @xmath17 @xmath14 0.1 ( left panel ) and @xmath13 @xmath14 0.7 ( right panel ) .
when weak wings are considered ( @xmath17 @xmath14 0.1 ) , the distribution is nearly uniform from 20 to 250 km s@xmath8 .
this is what is found by prochaska & wolfe ( @xcite ) .
however , when only the deepest part of the absorption is considered ( and we have given arguments showing that this part should be associated with the neutral phase ) , the distribution is peaked at small velocity broadenings .
most of the profiles have @xmath18 @xmath4 100 km s@xmath8 , while a few have @xmath18 @xmath9 150 km s@xmath8 .
it can be seen in fig .
[ pats_f ] ( lower left panel ) that the asymmetry parameter @xmath19 is correlated with the velocity broadening up to @xmath7 @xmath14 150 km s@xmath8 . for larger velocity broadenings
however this correlation disappears .
this approach complements the edge - leading test of prochaska & wolfe ( @xcite ; fig .
10 ) as it takes into account the correlation between asymmetry and velocity broadening that is expected in fast rotating thick - disk models . since it is the presence of systems with large velocity broadenings that mostly rules out the cdm model ( see fig
. 10 of prochaska & wolfe @xcite ) , we discuss below these systems individually : + @xmath20 _ q 0201 + 365 at @xmath10 = 2.462 : _ if we strictly follow our procedure and measure the velocity broadening of @xmath61608 at @xmath21 = 0.5 , we find @xmath22 @xmath14 600 km s@xmath8 .
this large velocity spread is due to a satellite of the main system , which has a column density log @xmath15 ( ) @xmath14 19 @xmath16 ( prochaska & wolfe @xcite ) . in this case , since the velocity separation is quite large , there is little doubt that we should consider that this satellite arises in a system that is _ not _ associated with the main damped component .
however this case clearly shows that the presence of satellites ( as expected in the case of merging processes ) could bias the discussion on the amount of rotation needed to explain the profiles . by restricting our consideration to the main system ,
the velocity broadening is @xmath7 @xmath14 225 km s@xmath8 .
the system does not show the edge - leading pattern . actually the ly@xmath0 line is fitted with 23 components ( prochaska & wolfe @xcite ) , none of which has log @xmath15 ( ) @xmath9 20 @xmath16 and four of which have log @xmath15 ( ) @xmath9 19 @xmath16 .
this system is definitively atypical .
+ @xmath20 _ q 0201 + 365 at @xmath10 = 1.955 : _ the @xmath61608 criterion gives @xmath7 @xmath14 230 km s@xmath8 .
here again there is no obvious edge - leading pattern .
even the ly@xmath0 line is poorly fitted by a damped component and the system resembles much more the lls system at @xmath10 = 2.325 , where strong components are spread over 200 km s@xmath8 , whereas the ly@xmath0 line does not show damped wings ( see prochaska & wolfe @xcite )
. + @xmath20 _ q 0528250 at @xmath10 = 2.811 : _ the system is at @xmath1 @xmath9 @xmath23 but is certainly not associated with the quasar ( warren & mller @xcite ; srianand & petitjean @xcite ) .
it has @xmath24
= 325 km s@xmath8 and the profile can be decomposed in two main sub - systems , each with no edge - leading structure .
we follow prochaska & wolfe ( @xcite ) and do not consider it as a typical damped system .
+ @xmath20 _ q 2233 + 131 at @xmath10 = 3.151 : _
this system is peculiar for several reasons .
djorgovski et al .
( @xcite ) have detected ly@xmath0 emission from this system .
the emission is redshifted from the main low - ionization absorption by about 270 km s@xmath8 ( lu et al .
@xcite ) .
since the @xmath61608 line does not reach @xmath13 = 0.7 at the resolution of the keck data and thus @xmath24 @xmath14 0 km s@xmath8 , we could only derive @xmath25 @xmath14 235 km s@xmath8 .
lu et al .
( @xcite ) have argued that this system is a case for which rotation is apparent in the absorption profile .
however it is clear from the @xmath61302 and @xmath61526 lines that this system is composed of three well detached sub - systems spread over 300 km s@xmath8 and centered at @xmath3270 , @xmath3190 and @xmath380 km s@xmath8 from the ly@xmath0 emission .
there is no edge - leading pattern as those emphasized by prochaska & wolfe ( @xcite ) .
instead the two sub - systems at @xmath3270 and @xmath3190 km s@xmath8 are equally strong and could both contribute to the damping wings .
it is interesting to note as well that the absorption does not correlate with the low - ionization absorptions .
the maximum of the absorption occurs at @xmath3120 km s@xmath8 , exactly in the gap between the low - ionization systems at @xmath3190 and @xmath380 km s@xmath8 .
this pattern is a clear example of what could be expected from a disturbed object made up of interacting sub - units .
it is qualitatively not different from the pattern seen in the @xmath1 = 2.8112 system toward q 0528250 .
rotational motions may be present in the sub - systems on velocity scales smaller than 100 km s@xmath8 .
+ from these comments , it seems that one has to be very careful when discussing the kinematics of systems with large velocity broadenings .
they often show the sub - systems that are expected if the objects are in the process of merging .
therefore the claim that the cdm model should be rejected has to be considered with caution . indeed using hydro - simulations , haehnelt et al .
( @xcite ) have reached the same conclusion .
even though these investigations are not yet complete , the overall picture that seems to be emerging favors the idea that dla systems are aggregates of dense knots with complex kinematics rather than ordered disks .
this does not rule out however the possibility that part of the kinematics could be due to rotation .
the correlation shown in fig .
[ pats_f ] ( lower left panel ) is suggestive of rotational motions in sub - systems on scales smaller than @xmath7 @xmath4 150 km s@xmath8 . finally ,
although the sample is certainly too small to draw any firm conclusion , it is interesting to note that the mean velocity broadening of at @xmath13 = 0.7 tends to decrease with redshift ( see fig .
[ patsf5z ] ; upper left panel ) _ if we do not include the peculiar systems mentioned in sect .
5.2_. if we split the sample in two sub - samples containing the same number of systems , the mean velocity broadening is 80 and 50 km s@xmath8 for @xmath10 @xmath4 2.2 and @xmath10 @xmath9 2.2 respectively .
this behavior is expected in cdm models in which disc galaxies form as gas cools and forms stars at the centres of the dark matter haloes ( e.g. kauffmann @xcite ) .
the examples of absorption profiles simulated by haehnelt et al .
( @xcite ) suggest that the comparison of the velocity profiles of high ( e.g. ) and low ( or ) ionization species may give clues on the nature of the objects associated with dla systems .
although fig . 6 of haehnelt et al .
( @xcite ) is drawn for illustrative purpose , it is unclear to us whether the simulations are able or not to reproduce observed individual cases .
contrary to what happens for at @xmath13 = 0.7 , there is no correlation between @xmath26 and @xmath27 at @xmath13 = 0.1 ( see fig .
[ pats_f ] ; lower right panel ) .
this is expected if the high - ionization phase has a much more disturbed kinematical field than the low - ionization phase ( see also turnshek et al .
@xcite ) and indeed , @xmath28 is most of the time larger than @xmath29 .
however , there is a trend for @xmath28 and @xmath29 both measured at @xmath21 = 0.1 to be correlated ( fig .
[ patsf5z ] ) .
the velocity differences between the maximum of the optical depth in and is most often smaller than 60 km s@xmath8 ( fig . [ patsf5z ] ) .
close inspection of the spectra shows that the kinematics of high and low - ionization species are similar .
exceptions are the @xmath1 = 2.827 system toward q 1425 + 606 , in which the main absorption corresponds to a high velocity sub - component , the @xmath1 = 3.151 system toward q 2233 + 131 , in which the strongest absorption is located in a gap between two low - ionization sub - components , and the @xmath1 = 1.776 system toward q 1331 + 170 , in which the profile is much more extended in velocity space than the low - ionization profiles .
such behaviors can be recognized in the simulated spectra of haehnelt et al .
( @xcite ) but should be investigated in more details .
it is interesting to note that the @xmath30/@xmath31 ratio seems to decrease with increasing redshift ( see fig .
[ patsf5z ] ) .
a very large value for this ratio is found for the peculiar @xmath1 = 2.844 system toward q 1946 + 769 which shows absorptions symmetrically placed on both sides of the low ionization absorption .
this general trend is consistent with kinematics to be more disturbed with decreasing redshift . |
since the discovery of pulsars there has been much discussion of observational evidence for decay of magnetic fields in neutron stars , as well as much theoretical work . as the reviews by lamb ( 1991 ) ,
chanmugam ( 1992 ) , and phinney & kulkarni ( 1994 ) indicate , there is at present no consensus on the question of whether or not magnetic fields in isolated neutron stars can decay significantly .
the general view has been that the electrical conductivity of matter in the cores of neutron stars is so high that the characteristic decay time for fields generated by electrical currents in the core is greater than the age of the universe .
recently pethick & sahrling ( 1995 ) showed that even if the conductivity in the core was small the shortest possible decay time is some two orders of magnitude longer than the decay time for configurations where the magnetic field is confined to the crust .
an incorporation of general relativistic effects , sengupta ( 1997 ) , further reduces the decay rate . still
, millisecond pulsars have typical field strengths in the range @xmath2 compared to isolated radio pulsars which have typically field strengths around @xmath3 and this indicates that during the spin - up phase the accretion process is reducing the field strength somehow .
millisecond pulsars are generally found in binary systems where the companion star is a white dwarf with a mass less than a solar mass , @xmath4 .
this system is called a low - mass - binary pulsar ( lmbp ) referring to the mass of the companion star . the progenitor to this system is thought to be the low - mass - xray binaries ( lmxb ) where a neutron star is accreting matter from a companion having a mass less than about 2 @xmath4 .
the accreting matter is spinning up the neutron star to millisecond periods .
for details concerning this process see reviews by phinney & kulkarni ( 1994 ) and bhattacharya & van den heuvel ( 1991 ) among others .
the evolution of the binary system after the neutron star is formed either by a supernovae or tidal capture , is assumed to occur on at least two time scales when the companion star is in radiative equilibrium . at first
the companion star is evolving on a nuclear time scale slowly filling its roche lobe .
when it has been filled up the matter overflows and falls onto the companion neutron star on a thermal time scale @xmath5 , see bhattacharya & van den heuvel ( 1991 ) for details . if lmxb s are the only progenitors to lmbp s the lifetime of the lmxb , or the accretion phase of the binary system , must be of order @xmath6
however , by using the amount of mass needed to be accreted to spin up the neutron star to a spin period @xmath7 one finds a time scale that ranges from @xmath8 .
this discrepancy suggests that there might other progenitors to the lmbp s .
for details see the review by phinney and kulkarni ( 1994 ) .
we will in this paper assume the accretion phase of progenitors to millisecond pulsars to last between roughly @xmath9 .
matter accreting onto a neutron star is expected to be disrupted by the magnetic field at some radius , @xmath10 , sometimes called the alfvn radius . beyond this bare description
there is no generally accepted view on how or where the matter attaches to the field lines and flows to the neutron star s surface , or on the interaction of the field with the matter outside @xmath10 , despite a large number of papers on the subject , see king ( 1995 ) . in the case of disk accretion
there are two main approaches to the problem . in one
( ghosh & lamb 1978 , 1979a , b ; kaburaki 1986 and wang 1987 ) the field is assumed to thread a large fraction of the disk because of kelvin - helmholtz instabilities . the other approach ( aly 1980 ; anzer & brner 1980 , 1983 ; scharlemann 1978 ) assumes the disk is a perfect conductor , completely excluding the field . in both approaches the matter is often assumed to leave the disk in i a narrow transition zone at the inner edge ( near @xmath10 ) , thereafter flowing along field lines to the neutron star .
most studies concerned with the magnetic field evolution _ inside _ the neutron star have assumed matter to accrete onto the surface spherically , e.g. fujimoto et al .
( 1984 ) , miralda - escud et al .
( 1990 ) , hereafter mir90 , urpin & geppert ( 1995 ) , and konar & bhattacharya ( 1997 ) . in this paper
i consider non - spherical accretion where matter is assumed to flow onto the polar caps in a column .
the cap is consequently heated up and the conductivity in the crust below the polar cap we estimate to be much smaller than the conductivity outside the accretion column .
the magnetic evolution in this scenario can be divided into two stages where in stage i the global decay rate is controlled by the conductivity outside the accretion column , @xmath11 . when the field has reached a value of about @xmath12
the accretion is spherical and the evolution enters the second stage .
here , the whole crust is being heated up and the conductivity is dominated by electron - phonon scattering . in this stage
the magnetic field decay time is a few orders of magnitude shorter than in stage i and compared to the earlier evolution the field is dissipating rapidly .
i argue that this could account for the fact that no binary pulsars have been found with a magnetic field less than @xmath13 .
a similar effect where the accreted flow is pushing the field lines has already been noticed by romani ( 1993 ) .
the paper is organized as follows : section 2 contains a brief description of the model and the basic equations . in section 3
we estimate the length scale for temperature change at the accretion column - normal crust boundary to be less than a crust thickness .
the time scale for the temperature to reach a stationary state is also shown to be much smaller than the magnetic field decay time scale .
therefore , we do not solve the energy equation explicitly but assume the conductivity as a function of angle to be close to a top - hat function at all times .
section 4 presents the numerical results for some initial depths of the magnetic field .
the purpose of the present paper is to investigate the effect of asymmetric accretion on the magnetic evolution in the _ crust _ of a neutron star .
we therefore ignore the complex interaction of matter and magnetic fields in the magnetosphere and for simplicity assume there is vacuum outside the star .
the basic model we are considering is an axi - symmetric spherical shell of outer radius @xmath14 and inner radius @xmath15 .
below @xmath15 there is a superconductor and outside the shell is vacuum .
i assume the equation of state in the shell to be given by baym et al .
furthermore , the magnetic field is assumed to vanish within the london penetration depth of the superconductor .
the magnetic field , @xmath16 , obeys the usual equation @xmath17 where we only include ohmic dissipation .
if one assumes the field to be initially poloidal , no torroidal field will appear and one can write the magnetic field in terms of the vector potential , @xmath18 in spherical coordinates .
one finds the following equation for @xmath19 : @xmath20 a detailed discussion of the boundary conditions used in this model can be found in sahrling ( 1996 ) . as a summary , the boundary conditions to be imposed are that , at the inner edge of the crust the field should be zero , no field diffuse into the core , and at the outer boundary of the crust the poloidal field should match onto the vacuum solution . to estimate conductivities we shall use the calculations of the electrical conductivity of matter in the crust of neutron star by urpin & yakovlev ( 1980 ) and raikh & yakovlev ( 1982 ) . to my knowledge , there exist no detailed studies of phonons in the inner crust of neutron stars , where nuclei are immersed in a neutron liquid .
however , one would expect the phonon spectrum to be qualitatively similar to that for a lattice of ions , provided one replaces the mass of an ion by the total mass of all nucleons in a cell of matter containing one nucleus . at high temperatures ,
the dominant source of electrical resistivity is scattering of electrons by phonons .
one gets from urpin & yakovlev ( 1980 ) the corresponding conductivity @xmath21 , where @xmath22 is the density in units of @xmath23 , @xmath24 is the mean molecular weight per electron , @xmath25 is the proton fraction of matter , and @xmath26 is the temperature in units of @xmath27 . for temperatures less than the debye temperature , @xmath28 ,
electron scattering is due chiefly to umklapp processes .
the corresponding conductivity is @xmath29 .
below a temperature @xmath30 k , umklapp processes are frozen out , and the conductivity rises very sharply , since only normal processes contribute , and the electrical conductivity due to phonon scattering is @xmath31s@xmath32 . however , at temperatures below @xmath33 , impurities are likely to be the dominant scatterers of electrons .
one can see this from the estimate for the impurity contribution , @xmath34 ( @xcite ) . here
@xmath35 is the mean square deviation of the atomic number from its average value .
the impurity parameter @xmath35 is not well known , but the estimate of flowers & ruderman ( 1977 ) suggests @xmath36 , to be a reasonable lower limit .
the problem of accretion onto neutron stars and the interaction with the magnetosphere is a major research area in modern astrophysics and many interesting and challenging problems remain to be solved , see for example lamb ( 1991 ) , michel ( 1991 ) , white et al .
( 1995 ) for interesting discussions of this and related phenomena .
i assume here the accreting matter to fall upon the neutron star in an accretion column whose size is given by the last open field line before the accretion disk . for simplicity
, i also assume there be no `` slipping '' between field lines as argued by arons & lea ( 1980 ) .
the basic idea ( lamb et al .
1973 and shapiro & teukolsky 1983 ) is that matter is captured by some instability mechanism at @xmath10 where the kinetic energy density of the accreting matter is equal to the magnetic energy density , @xmath37 , where @xmath38 is the surface magnetic field strength in units of @xmath39 gauss , @xmath40 is the accretion rate in units of @xmath41 , @xmath42 is the stellar radius in units of @xmath43 cm , and @xmath44 is the mass of the neutron star .
matter is then channeled by the field lines onto the neutron star surface in a column with an opening angle @xmath45 .
the area under the accretion column is @xmath46 .
the angle @xmath45 is given by those dipole field lines that would , in the absence of accretion , have penetrated beyond @xmath10 .
the field lines for a dipole field are defined by @xmath47 , so the base of the last undistorted field line which would close inside @xmath10 lies at an angle @xmath48,\ ] ] for a detailed derivation see e.g. shapiro & teukolsky ( 1983 ) . as the field strength decreases due to
ohmic losses the opening angle gets larger and larger
. eventually , the field strength will be so low the accretion is expected to be spherical since then @xmath49 .
we have @xmath50 $ ] where @xmath51 is the value of the magnetic field when the accretion is spherical .
this is of course a very simplified picture of the accretion process , but it will nevertheless provide some insight as to how the crust affects the surface magnetic field .
this section discusses some relevant order of magnitude estimates .
although there are many heat sources in an accreting neutron star we will use hydrogen burning in a thin shell at @xmath52 , see mir90 , as the major one .
mir90 calculated models suppressing all other heat sources and came up with temperature differences of a about a factor @xmath53 .
the density where the accreted material starts to spread laterally , @xmath54 is shown to be about an order of magnitude higher than the base of the hydrogen burning shell , @xmath55 .
we also estimate typical thermal length scales to be shorter than the crust thickness .
these estimates indicate the heat source in our model to have a size of about @xmath56 and located at @xmath57 .
further , it is shown that the time for the temperature distribution to reach a steady state is much shorter than typical magnetic field decay time scales .
the kinetic energy of the accreting matter is turned into heat in the upper parts of the atmosphere . after being
thermalized the matter continues to flow downwards at sub sonic speed .
it will follow the magnetic field lines down to a depth where the local magnetic force equals the local pressure gradient . at this point
the flow is presumed to spread laterally , see bildsten & brown ( 1996 , 1997 ) .
they found @xmath58 , where @xmath59 and @xmath60 is a pressure scale height .
if we assume the pressure to be dominated by relativistic , degenerate electrons we find @xmath61 . as the accreted material flows down along the field lines its density will increase which at certain thin shells will cause nuclear reactions to occur which releases heat . for a thorough discussion of these matters
see for example brown & bildsten ( 1997 ) , bildsten et al .
( 1993 ) , and mir90 .
the base of the hydrogen burning shell @xmath62 for the spherical accretion case , see mir90 . in our model ,
the size of the polar cap @xmath63 so one can expect @xmath55 to be similar to the spherical calculation .
we see for typical values of the relevant parameters @xmath64 .
however , it is also clear that as the field strength is reduced @xmath54 is decreasing and at some some value of @xmath65 , @xmath54 may be less than @xmath55 and the flow might start to spread before hydrogen ignites .
the resulting temperature and density structure for that case must ultimately be answered by a thorough magneto - hydrodynamical calculation , and in this paper i will use a simple parametrization . assuming the energy generation in the accretion column to be dominated by hydrogen burning at the @xmath66-decay limited value of the `` hot '' cno cycle we get from mir90 a luminosity @xmath67 , where @xmath68 is the heavy element content , or a flux @xmath69 . to find a typical length scale for temperature variations , which we denote by @xmath70 , one can equate this flux with the thermal heat flux @xmath71 , where @xmath72 is the thermal conductivity calculated by urpin & yakovlev ( 1980 ) for the case where the @xmath73 . the radiative contribution to
the heat flux is negligible for the temperatures and densities of interest here , mir90 and mszros ( 1992 ) .
we get @xmath74 where i have normalized the density to @xmath75 which is roughly the value of @xmath55 .
@xmath76 is the area of the accretion cap in units of @xmath77 and @xmath78 .
we see that @xmath79 is typically smaller than a crust thickness , @xmath80 and for some parameter values it can be much smaller . when @xmath81 or @xmath82 the temperature structure in the crust is spherical which has been studied elsewhere , see mir90 .
here , i will focus on the intermediate case , where @xmath83 . to get an estimate of the temperature structure below and around the accretion cap
we can simply note that as the conductivity increases inwards the thermal lengthscale also increases so for a given temperature change , @xmath84 , we need to go a distance @xmath85 radially which is greater than the corresponding horizontal distance @xmath86 where @xmath87 . in other words
, the hydrogen burning shell will produce a temperature structure that has an ellipsoidal or , as i will assume in this paper , columnar shape . in the estimate of
@xmath79 we have simply assumed the relevant area to be @xmath88 for all densities , and so ignored the heat flowing through the sides of the column .
however , since the size of the polar cap is of the same order as the crust thickness this correction is likely to be small . given the uncertainty of the horizontal lengthscale and the size of the hot accretion cap i choose to parameterize @xmath79 in this paper , see section 4 .
the time scale for reaching a steady state is now given by , @xmath89 where we have assumed @xmath90 which holds below @xmath91 .
@xmath92 is the specific heat capacity of the crust , see e.g. shapiro & teukolsky ( 1983 ) and @xmath93 is the atomic number .
the magnetic field decay time is given by @xmath94 which with @xmath95 is @xmath96 . when the accretion is spherical @xmath97 and @xmath98 .
we see that @xmath99 when the conductivity is dominated by either phonon or impurity scattering . in the estimates of the field decay times
we have normalized the values of the parameters to those of the inner crust .
the temperature will thus reach a steady state long before the magnetic field has changed significantly .
given these estimates it is likely that the temperature structure in the crust of a neutron star , accreting in a column , is such that inside the accretion column the temperature resembles the calculation by mir90 and by simply extrapolating their result we find , @xmath100 outside the column the temperature is much lower .
precisely how much lower remains to be seen by detailed calculations but motivated by the previous estimates of @xmath79 we will in this paper assume it to be low enough for impurity scattering to be the dominant source of electrical conductivity @xmath95 .
i use for the conductivity , @xmath101 where @xmath102 , @xmath103 and @xmath104 is a correction used to ensure that @xmath105 when the accretion is uniform over the whole sphere @xmath49 . to see how the size of the horizontal thermal length scale affect the solution i use a simple parameterization of @xmath106 where @xmath66 is assumed to be independent of @xmath107 and in this paper range from @xmath108 .
it is worth pointing out that if the different parameters are such that @xmath109 initially , @xmath79 will increase as the magnetic polar cap size increases and could eventually be such the whole crust is heated up .
the spatial evolution of the magnetic field in the crust will then be different from this calculation .
however , the _ temporal _ evolution , will be very similar , that is the timescale for the accretion to be completely spherical is the same as we find here , and also the subsequent rapid field decay .
this section describes numerical solutions to equations ( [ eq : psi1 ] ) and ( [ eq : conductivity ] ) with two different initial conditions .
first , i choose a configuration where the field is penetrating the whole crust , @xmath110 . in the second case
the field was initially confined to @xmath111 . in both cases
the initial surface dipole strength was @xmath39 gauss .
the results are combined into three sections where in the first , section 4.1 , the field strength is initially penetrating the whole crust .
this is followed by section 4.2 where the field fills only the outer crust @xmath112 initially .
the numerical calculations show the decay time scale can be divided into two stages where in stage i the decay time scale is given by @xmath113 and when the accretion becomes spherical it enters the second stage where the decay rate increases to @xmath114 .
the calculations also show that starting from a dipole , the contribution from higher order surface multipoles is at the most @xmath115 to the dipole mode in agreement with arons ( 1993 ) . in section 4.3
i explore different values of the horizontal length scale by varying @xmath66
. equations ( [ eq : psi1 ] ) and ( [ eq : conductivity ] ) was solved with the finite - difference code described elsewhere , sahrling ( 1996 ) . for the impurity parameter
i chose four values @xmath116 and @xmath117 which span the likely range of values , see flowers & ruderman ( 1977 ) , and is constant throughout the crust .
the model i am considering here is an @xmath118 neutron star with the equation of state given by wiringa et al .
( 1988 ) and baym et al .
( 1971 ) , with radius @xmath119 .
the equation of state for neutron stars is a debated issue but this choice represents an average stiffness of the range of models discussed . at @xmath120 the field
was set to be dipolar , @xmath121 .
since the conductivity below the accretion cap @xmath122 and @xmath123 is smaller than the conductivity outside , the field will initially change more rapidly there producing higher - order multipoles .
generally , it was found the dependence of the various runs on the impurity parameter @xmath35 is simple .
let us denote by @xmath124 the time scale for runs with a particular @xmath35 .
then @xmath125 which given our approximations is to be expected . from the observational constraints discussed in the introduction we can infer timescales for accretion
lies roughly between @xmath126 depending on such things as mass of the companion star and the spin period @xmath7 .
the numerical results of our model indicates that the value of @xmath35 has to be greater than about 0.1 to give a significant decay of the magnetic field .
the field strength was chosen to be @xmath39 gauss at the stellar surface initially .
this high field is consistent with the pattern of high - mass x - ray binaries ( hmxbs ) , in which the x - rays originate in a small steady hot patch , see lyne & graham - smith ( 1990 ) .
there is however one lmbp system known , psr 0820 + 02 , with a similar high field strength , so this initial condition can also be relevant for lmxb s . to calculate @xmath45 from equation ( [ eq : tetacap ] ) we use a constant @xmath127 .
the parameter @xmath128 for all runs discussed here .
figure 1 shows the evolution of the various surface magnetic multipoles , normalized to the initial value @xmath65 .
it shows the evolution for a crust with @xmath129 .
this high impurity level results in very low conductivity and thus the most rapid field decay among the cases considered . changing the value of @xmath35 resulted only in a change in time scale of a factor @xmath130 . as can be seen from the figure high - order multipoles are initially stronger than multipoles @xmath131 but as time progresses they disappear more and more and the lower ones start to dominate .
the contribution of @xmath131 compared to the dipole is about 10 % at the time the accretion is spherical , which happens when @xmath132 and the field evolution enters stage ii .
we saw earlier that in our model this value varies with the square root of the accretion rate .
one has to conclude that if this mechanism is responsible for there being no pulsars with a magnetic field lower than @xmath133 gauss the accretion rate for the majority of the millisecond pulsars during their recycling phase will have to have been , on average , greater than roughly @xmath134 .
observationally it is hard to pin point the precise accretion rate but assuming the luminosity to be due solely to the infalling matter one finds accretion rates ranging from @xmath135 phinney & kulkarni ( 1994 ) .
if one instead starts out with a field buried below @xmath111 one finds the result presented in figure 2 .
there is an initially rapid decay because the field diffuses into the interior of the star on a time scale , @xmath113 calculated at @xmath136 .
the field starts to penetrate the whole crust at a time scale given by the conductivity of the inner crust .
this is all what can be expected from dimensional analysis of the induction equation .
the @xmath131 multipole strengths are in this case also about 10 % of the dipole when the accretion is spherical . to investigate the sensitivity of the solution to different horizontal thermal length scales we used three different values of the parameter @xmath137 . to avoid numerical problems i chose not to go below @xmath138 .
the initial field was assumed to penetrate the whole crust and have an initial surface strength of @xmath39 gauss .
the results are shown in figure 3 a),b ) .
these plots should be compared to fig .
1 . as can be expected , the higher order multipoles are less significant when @xmath66 increases .
other than that the overall evolution is very similar for all values of @xmath66 considered .
we have examined the influence of asymmetric accretion on the magnetic field evolution of a neutron star .
the temperature structure in the crust resulting from the accretion was roughly estimated and found to vary more rapidly than the crust thickness around the edge of the accretion cap . the time scale for reaching a steady state
was shown to be shorter than the time scale for magnetic field decay .
therefore , instead of doing a full blown calculation of heat conduction coupled with the magnetic field evolution we used a simple smoothed top - hat function for the temperature structure and consequently for the conductivity .
the global field decay time is roughly the one given by the largest conductivity in the crust , which occurs outside the accretion cap , @xmath139 .
however , when the field strength is down to roughly @xmath133 gauss the accretion is spherically symmetric and we have @xmath140 and so the decay rate is much faster than initially which could provide a clue why no binary pulsars are known with field strengths less than @xmath133 gauss .
romani ( 1993 ) discussed a similar effect and found a threshold close to ours .
we also found the asymmetric accretion resulted in some higher order surface magnetic multipoles and the size of these were shown to be at most 30 % of the surface dipole field .
* acknowledgements :* i would like to thank dong lai , lars bildsten and edward brown for useful discussions .
dong lai is especially thanked for a careful reading of the manuscript .
this work was also supported in part by the u. s. national science foundation under grants nsf ast93 - 15133 and ast94 - 14232 , by nasa under grant nagw-1583 , and by the swedish natural science research council .
aly , j.j . , 1980 , a&a , 86 , 192 anzer u. , brner g. 1980 , a&a , 83 , 133 anzer u. , brner g. 1983 , a&a , 122 , 73 arons , j , lea , s. 1980 , , 235 , 1016 arons , j. 1993 , , 408 , 160 baym , g. , bethe h. a. , pethick , c.j . , 1971 , nuclear physics a , 175 , 225 bhattacharya , d. , van den heuvel , e. p. j. 1991 , phys . rep . 203 , 1 brown , e. f. , bildsten , l. 1997 , to appear in apj bildsten , l. , brown , e. f. 1996 , , 477 , 897 bildsten , l. , salpeter , e. e. , wasserman , i. 1993 , , 408 , 615 chanmugam , g. 1992 . , 30 , 143 flowers , e. , ruderman , m. a. 1977 , , 215 , 302 fujimoto , m. y. , hanawa , t. , iben , i. , jr . ,
richardson , m. b. 1984 , , 278 , 813 ghosh , p. , lamb , f. 1978 , l , 223 , l83 ghosh , p. , lamb , f. 1979a , , 232 , 259 ghosh , p. , lamb , f. 1979b , , 234 , 296 kaburaki , o. 1986 , , 220 , 321 king , a. , 1995 in x - ray binaries , eds .
lewin w.h.g . , paradijs van , j. , van den heuvel , e.p.j .
( cambridge : cambridge university press ) konar , s. , bhattacharya , d. 1997 , , 284 , 311 lamb , f. k. , pethick , c. j. , pines , d. 1973 , , 184 , 271 lamb . f. 1991 , in frontiers of stellar evolution , ed .
d. lambert ( san francisco : astronomical society of the pacific ) , 299 lyne , a.g.,graham - smith , f. 1990 , pulsar astronomy , ( cambridge : cambridge university press ) mszros , p. 1992 , high - energy radiation from magnetized neutron stars , ( chicago : university of chicago press ) michel , f. c. 1991 , theory of neutron stars magnetospheres , ( chicago : university of chicago press ) miralda - escud , j. , haensel , p. , paczyski , b. ( mir90 ) 1990 , , 362 , 572 pethick , c.j . , sahrling , m. 1995 , l , 453 , l29 phinney , s.,kulkarni , s. 1994 , , 32 , 591 raikh , m .
e. and yakovlev , d. g. 1982 , , 87 , 193 romani , r. w. 1993 , in isolated pulsars , eds . k. a. van riper , r. epstein , c. ho ( cambridge : cambridge university press ) sahrling , m. 1996 , submitted to the astrophysical journal scharlemann , e.t .
1978 , , 219 , 617 sengupta , s. 1997 , to appear in the astrophysical journal letters shapiro , s. , teukolsky , s. 1983 , the physics of compact objects , ( new york : wiley ) urpin , v. a.,geppert , u. 1995 , , 275 , 1117 urpin , v. a. , yakovlev , d. g. 1980 , , 24 , 303 wang , y. m. 1987 , a&a , 183 , 257 white , n.e . , nagase f. , parmar a.n .
1995 , in x - ray binaries , eds .
lewin w.h.g . , paradijs van , j. , van den heuvel , e.p.j .
( cambridge : cambridge university press ) wiringa , r.b . , fiks , f. , fabrocini , a. 1988 , physical review c , 38 , 1010 the evolution of the various surface magnetic multipoles , normalized to the initial value @xmath141 with @xmath129 and @xmath128 .
initially the filed penetrated the whole crust .
high - order multipoles are initially stronger than low - order ones but as time progresses they disappear more and more and the lower ones start to dominate .
the dips in the curves corresponds to times when the multipoles changes sign .
the contribution of @xmath131 compared to the dipole is about 10 % at the time when the accretion is spherical .
the early part of the evolution from @xmath120 is not shown here since that describes the initial transients of the field due to the choice of initial conditions @xmath142 . | we consider magnetic field evolution of neutron stars during polar - cap accretion .
the size of the polar cap increases as the field decays , and is set by the last open field line before the accretion disk . below the polar cap we find the temperature to be so high that electron - phonon scattering dominates the conductivity . outside the polar cap region ,
the temperature is such the conductivity is dominated by temperature independent impurity scattering which can be a few orders of magnitude larger than the electron - phonon conductivity .
the time - scale for field decay is therefore initially given by impurity scattering dominated conductivity .
when the field strength has been reduced to @xmath0 the accretion is spherical and the time scale for field decay is given by the smaller electron - phonon scattering conductivity .
the field strength is now reduced rapidly compared to before and this could be a reason for there being no pulsars known with field strengths below @xmath1 .
we also investigate the evolution of multipoles at the neutron star surface .
we find that contribution from higher - order multipoles are at most 30 % to that of the dipole mode . |
synthesis and investigation of very neutron - rich nuclides are of exceptional importance to advance our understanding of nuclear structure and properties at the extreme isospin limit of the nuclear landscape .
many of these nuclides play a key role in stellar nucleosynthesis , especially in the r - process .
the production and separation of these nuclides represent an essential ingredient in regards to current or future radioactive beam facilities ( see e.g. @xcite ) .
fission or spallation are prolific ways to generate a variety of neutron - rich nuclides ( see e.g. @xcite ) . in addition ,
projectile fragmentation of n - rich beams @xcite at high or intermediate energy ( well above the fermi energy ) has been efficiently employed to produce radioactive beams of n - rich nuclei .
the so - called `` cold '' projectile fragmentation has been described by abrasion / ablation models which provide a phenomenological basis for extrapolation to even more exotic n - rich species @xcite .
based on available fragmentation data , a detailed parametrization of the fragmentation cross sections from high - energy reactions has been developed @xcite and is commonly used for planning of radioactive beam experiments . in high or medium energy projectile fragmentation reactions ,
the production of n - rich rare isotopes is based on a more or less `` clean - cut '' removal of protons from the projectile . in such reactions ,
the target has almost no effect on the production cross sections , apart from a geometric factor .
one - neutron pick - up products are produced with very small cross sections @xcite .
in contrast to high - energy reactions , the effect of the target neutron - to - proton ratio ( n / z ) in n - rich rare isotope production has been shown in multinucleon transfer reactions close to the coulomb barrier .
for example , large cross sections for several neutron pick - up channels , along with proton stripping channels have been observed in the reaction of @xmath1ni projectiles with @xmath3u above the coulomb barrier @xcite .
the effect of the projectile and target
n / z in the production of projectile - like fragments from @xmath4s ( e / a=620 mev / nucleon ) on @xmath5c and @xmath6au has recently been reported in @xcite , following earlier work in this energy range @xcite .
deep inelastic collisions around the coulomb barrier have also been used to produce and study n - rich rare earth isotopes at high spin states @xcite . around the fermi energy ( 2040 mev / nucleon ) a number of n - rich yield measurements of heavy projectiles ( e.g. @xmath7ar , @xmath0kr ) exist in the literature @xcite , but in these measurements no production cross sections have been reported .
cross sections have been reported for the reaction 70 mev / nucleon @xmath0kr + @xmath8al @xcite , but the n - rich side of the fragment distribution was not covered . in the present work ,
we have performed a systematic high - resolution spectrometric study of the production cross sections and the velocity distributions of projectile - like fragments from the reaction 25 mev / nucleon @xmath0kr + @xmath1ni with particular attention on the n - rich nuclides . in this study , we observed enhanced production cross sections of n - rich fragments near the projectile and the formation of several neutron pick - up products , along with proton stripping products .
the paper is organized as follows : section 2 describes the experimental setup , the measurements and the data analysis procedures . in section 3 , the experimental results are presented and compared with model predictions . finally ,
in section 4 , a summary and conclusions are given .
the present study was performed at the cyclotron institute of texas a&m university . a 25 mev
/ nucleon @xmath0kr@xmath9 beam from the k500 superconducting cyclotron , with a typical current of @xmath105 pna , interacted with an isotopically enriched ( 98% ) @xmath1ni target of thickness 4 mg/@xmath11 .
the reaction products were analyzed with the mars recoil separator @xcite .
the primary beam struck the target at 0@xmath12 relative to the optical axis of the spectrometer .
the direct beam was collected in a small square faraday cup approx .
30 cm after the target , blocking the angular range 0.01.0@xmath12 .
the fragments were accepted in the remaining angular opening of mars : 1.02.7@xmath12 ( the angular acceptance of mars is 9 msr @xcite ) .
this angular range lies inside the grazing angle of 3.6@xmath12 @xcite for the present reaction .
an al foil ( 1 mg/@xmath13 ) was positioned after the faraday cup to reset to equilibrium the ionic charge states of the reaction procucts .
mars optics @xcite provides one intermediate dispersive image and a final achromatic image ( focal plane ) and offers a momentum acceptance of 4% . at the focal plane , the fragments were collected in a large area ( 5@xmath145 cm ) three - element ( @xmath15e@xmath16 , @xmath15e@xmath17 , e ) si detector telescope .
the @xmath18e@xmath16 detector was a position - sensitive si strip detector of 63 @xmath19 m thickness whereas the @xmath18e@xmath17 and the e detector were single - element si detectors of 150 and 950 @xmath19 m , respectively .
the position information from the @xmath15e@xmath16 strips provided a continuous monitoring of the focusing and collection of the fragments at the various settings of the separator .
time of flight was measured between two ppacs ( parallel plate avalanche counters ) @xcite positioned at the dispersive image and at the focal plane , respectively , and separated by a distance of 13.2 m. the ppac at the dispersive image was also x y position sensitive and used to record the position of the reaction products . the horizontal position , along with nmr measurements of the field of the mars first dipole , was used to determine the magnetic rigidity @xmath20 of the particles .
thus , the reaction products were characterized by an event - by - event measurement of de / dx , e , time of flight , and magnetic rigidity .
the response of the spectrometer / detector system to ions of known atomic number z , mass number a , ionic charge q and velocity was calibrated using low intensity primary beams of @xmath7ar , @xmath21ca and @xmath0kr at 25 mev / nucleon . to cover the n / z and velocity range of the fragments , a series of measurements was performed at overlapping magnetic rigidity settings in the range 1.62.0 tesla - meters .
the determination of the atomic number z was based on the energy loss of the particles in the first @xmath22 detector @xcite and their velocity , with a resulting resolution ( fwhm ) of 0.5 z units for near - projectile fragments . the ionic charge @xmath23 of the particles after the al stripper ,
was obtained from the total energy e@xmath24=@xmath18e@xmath25+@xmath18e@xmath26+e , the velocity and the magnetic rigidity according to the expression : @xmath27 where e@xmath24 is in mev , b@xmath28 in tm , @xmath29 and @xmath30 .
the measurement of the ionic charge q had a resolution of 0.4 units ( fwhm ) . since the ionic charge must be an integer , we assigned integer values of q for each event by putting windows ( @xmath31 ) on each peak of the q spectrum . using the magnetic rigidity and velocity measurement ,
the mass - to - charge a / q ratio of each ion was obtained from the expression : @xmath32 combining the q determination with the a / q measurement , the mass a was obtained as : @xmath33 ( q@xmath34 is the integer ionic charge determined as above ) with an overall resolution ( fwhm ) of about 0.6 a units ( fig .
[ mass ] ) . combination and appropriate normalization of the data at the various magnetic rigidity settings of the spectrometer provided fragment distributions with respect to z , a , q and velocity .
correction of missing yields caused by charge changing at the ppac ( positioned at the dispersive image ) was performed based on the equilibrium charge state prescriptions of leon et .
( the overall data reduction procedure was similar to that followed in earlier work on @xmath6au fragmentation and was described in detail in @xcite . )
the distributions were subsequently summed over all values of q. it should be pointed out that the resulting distributions in z , a and velocity are the fragment yield distributions in the reaction angle interval 1.02.7@xmath12 in the magnetic rigidity range 1.62.0 tm . fig .
[ mass ] shows , as an example , the mass spectrum of germanium ( z=32 ) isotopes in full resolution .
the gross features of the distributions of projectile fragments from the present reaction are described in fig .
[ gross ] .
the detailed mass distributions of elements z=35 to z=30 are presented in fig .
[ z1 ] . before further discussion of the data
, we will give an outline of the calculations we performed for this reaction using a phenomenological model appropriate for this energy regime .
the primary interaction stage was modeled with the deep inelastic transfer code of tassan - got and stephan @xcite in which stochastic nucleon exchange was assumed for the orbital angular momentum range @xmath35=100520 .
this dit model has been successfully applied to describe the primary interaction stage in studies of projectile multifragmentation of @xmath36si on @xmath37sn around the fermi energy @xcite .
events corresponding to trajectories in which the projectile target overlap exceeded 3 fm were rejected . following the creation of the primary fragments by the dit mechanism ,
the statistical de - excitation of the excited primary fragments was simulated using gemini @xcite .
this statistical deexcitation code uses monte carlo techniques and the hauser - feshbach formalism to calculate the probabilities for fragment emission with z@xmath382 .
heavier fragment emission probabilities are calculated using the transition state formalism of moretto @xcite . in the gemini calculations
, we used lestone s temperature dependent level density parameter @xcite , a fading of shell corrections with excitation energy and we enabled imf emission . each partial - wave distribution was appropriately weighted and combined to give the overall fragment a , z and velocity distributions .
the results of the dit / gemini calculation were also filtered by the angular and b@xmath39 acceptance of the spectrometer .
the predictions of this calculation are compared with the present data in the following paragraphs . in fig .
[ gross]a the mass yield curve is presented .
the measured data , normalized for beam current and target thickness are given in mb and presented as open circles .
the result of the dit / gemini calculation , filtered by the spectrometer acceptance is given by the dashed line , whereas the full line gives the total ( unfiltered ) yield .
a comparison of the measured yields ( open symbols ) to the calculated filtered yields ( dashed line ) shows excellent agreement for the heavier fragments ( a@xmath4065 ) .
the discrepancies for lower mass fragments are mainly due to incomplete coverage of the measured data for this mass range . using the ratio of filtered to
unfiltered calculated yield for each mass , correction factors ( whose magnitude is inferred from fig . [ gross]a ) for the acceptance of the spectrometer were obtained as a function of mass and were applied to the measured yield data to obtain the total yield , given by the full circles in fig .
[ gross]a .
the systematic uncertainty in the extraction of absolute cross sections by this procedure is estimated to be about 40% ( fwhm ) .
the correction factors were also employed to obtain total isotope production cross sections ( fig .
[ z1 ] ) from the measured yields . a comparison of the extracted total isobaric cross sections ( full circles ) to the calculated unfiltered yields ( full line ) again shows fair agreement , except for the lower mass fragments ( a@xmath4165 ) . in fig .
[ gross]b , the measured yield distributions as a function of z ( relative to the line of @xmath42 stability , z@xmath43 ) and a are presented as contour lines . the line of stability is calculated as : z@xmath44 = a/(1.98 + 0.0155a@xmath45 ) @xcite .
the calculated values from dit / gemini are shown as a thick full line ( without acceptance cut ) and as a thick dashed line ( with spectrometer acceptance cut ) .
the thin dashed line is from the epax parametrization @xcite of relativistic fragmentation data and is given here for comparison . from this figure
we see that very neutron - rich nuclides ( up to 4 z units away from stability for isobars near the projectile ) are produced .
the dit / gemini calculation describes reasonably well the experimental data . finally , the locus of the relativistic fragmentation data is towards more proton - rich fragments than the data in this energy regime . finally , in fig .
[ gross]c , the velocity vs. mass distributions are given .
the data are again shown as contours .
the thick full line is from the dit / gemini calculation without acceptance cut and the dashed line is with acceptance cut .
this mass resolved velocity distribution exhibits correlations characteristic of quasi - elastic interactions ( for near projectile fragments ) and deep inelastic interactions ( for fragments further away from the projectile ) as revealed by the continuous fragment ridge with monotonically decreasing velocities . for lower masses ( a@xmath4165 )
the fragment velocities seem to increase with decreasing mass .
this is due to the fact that the spectrometer settings were such that only forward moving fragments coming from decay of highly excited primary products were observed .
in addition , comparing the filtered dit / gemini calculations with the data , we see that the calculation is able to describe well the observed velocity distributions in the whole mass range . in summary ,
based on the comparisons presented in fig .
[ gross ] , we can state that the dit / gemini calculation is able to provide a satisfactory quantitative description of the observed gross distributions .
also , as we will see below , it does a fair job in predicting the absolute values of the production cross sections ( except for the very n - rich isotopes , as will be pointed out ) . in fig .
[ z1 ] , the mass distributions for elements z=3530 are presented . in this figure ,
the full circles are the present data , corrected for the spectrometer acceptance as discussed earlier .
the open squares are the predictions of the dit / gemini calculation .
the dotted lines are the predictions of the epax parametrization @xcite of relativistic fragmentation cross sections and are plotted here for comparison .
note that , in high - energy fragmentation , nucleon - pickup products are not produced or , at best , are highly suppressed compared to lower energy peripheral collisions @xcite .
as we see , neutron - rich nuclides are produced with large cross sections .
apart from proton - removal products , neutron pick - up fragments are produced .
for example , for the case of selenium ( z=34 ) , which corresponds to the removal of 2 protons from the projectile , up to 4 neutrons are seen to be picked - up to produce @xmath46se ( with cross section of 0.3 @xmath19b ) .
similarly , for germanium ( four - proton - removal from the projectile ) , up to two neutrons are picked - up to give @xmath47ge ( cross section 0.4 @xmath19b ) . for near - projectile elements , an enhanced production of neutron rich isotopes
is observed relative to the expectations of the epax parametrization .
also , the dit / gemini calculation , while able to describe rather well the n - deficient and the central part of the distributions , fails to describe the n - rich sides of the distributions for elements above zinc ( z=30 ) .
since this enhancement is observed for near projectile elements , it should be a result of very peripheral collisions , where the nucleon exchange is restricted near the surface of the projectile and the target . to further examine this statement , we present in fig .
[ vel ] the velocity distributions for elements z=3530 .
these distributions have two components : a narrow one peaking close to the beam velocity ( quasielastic ) and another wider component peaking at lower velocity ( deep inelastic component ) .
the quasielastic component progressively decreases for fragments with lower z. as expected , the production of the most neutron - rich isotopes is associated with the quasielastic component ( also verified in two - dimensional correlations of a with velocity ) .
using velocity vs impact parameter correlation from the dit / gemini calculation , we found that these peripheral events correspond to a projectile
target overlap not exceeding 11.5 fm .
qualitatively , the enhanced production of neutron rich fragments from @xmath0kr may be understood by considering the peripheral character of the collisions and the particular structure of the projectile and the target .
the projectile has a compact neutron distribution ( n=50 closed shell ) and , reversely , the target has a compact proton distribution ( z=28 closed shell ) and a more diffuse neutron distribution ( n=36 , partially filled neutron shell with 8 neutrons above the closed shell n=28 ) . in peripheral collisions ,
an overlap of the @xmath0kr projectile surface with the neutron - rich @xmath1ni surface ( neutron `` skin '' ) will lead to a local nucleon redistribution and n / z equilibration , favoring the production of more neutron - rich fragments near the projectile . for larger projectile
target overlaps this effect washes out and the observed fragments have more or less cross sections as described by dit / gemini .
interestingly , they are in general agreement with epax , despite the vastly different primary ineraction mechanism , indicating that such products result from long evaporation chains of highly excited primary nuclei .
it should be noted that in the dit code , the nuclei are assumed spherical with homogeneous proton and neutron density distributions .
incorporating realistic proton and neutron density distributions in the dit code may improve the capability to describe peripheral collisions in projectile
target combinations where the neutron proton profile of nuclear surface can play a role , as in the present case of the @xmath0kr+@xmath1ni system . from a practical standpoint , the large production cross section of neutron - rich nuclides , and
more importantly the neutron pick - up possibility can render these reactions a useful route to produce extemely neutron - rich nuclides .
it may be noted that neutron pick - up cross sections are also large in the case of multinucleon transfer reactions close to the coulomb barrier @xcite . however ,
these reactions have very wide angular , velocity and ionic charge distributions , in addition to the necessity of rather thin targets ( around 1 mg/@xmath13 ) . around the fermi energy ,
however , inverse kinematic reactions have angular and ionic charge state distributions that can be efficiently dealt with using a large acceptance spectrometer ( e.g. mars in the present study ) .
also the energies are high enough to allow moderately thick targets ( 1030 mg/@xmath48 . using the present cross sections
, we can make estimates of rare isotope production rates from intense beams at this energy regime . as examples , we present in table i the cross sections and production rates for the most n - rich nuclides of se ( z=34 ) and ge ( z=32 ) .
the experimental cross sections of this work are given in the second column , while the the third column gives the cross sections predicted by the dit / gemini calculation .
the fourth column gives the cross sections measured in the reaction @xmath0kr(500 mev / nucleon ) + be @xcite .
we observe , as already discussed , the inability of the present simulation to reproduce the measured cross sections of the most n - rich nuclides .
also , we note that in relativistic peripheral collisions , up to one neutron can be picked up from the target with very low cross section . the rates given in the last column of the table
were calculated using the measured cross sections and assuming a beam of 100 pna @xmath0kr at 25 mev / nucleon striking a 20 mg/@xmath11 @xmath1ni target .
such yields of very neutron rich isotopes may enable a variety of nuclear structure and nuclear reaction studies in the fermi energy regime .
these rare isotopes may be separated in flight with a large acceptance separator or can be stopped and collected in a gas cell with the possibility of subsequent charge breeding and reacceleration @xcite . finally , another interesting possibility is the use of such reactions as a second stage in two - stage rare isotope production schemes .
for example , a beam of @xmath49kr from an isol facility can be accelerated around the fermi energy and subsequently strike a @xmath1ni target to produce very n - rich nuclides that may be separated and studied in flight . to estimate the rates of such reaction products the present cross sections can be used as a first approximation .
however , a quantitative prediction will be possible after improving our present description of these peripheral collisions between n - rich nuclei .
in the present study , the yields and velocity distributions of projectile - like fragments from the reaction of 25 mev / nucleon @xmath0kr + @xmath1ni were measured using the mars recoil separator at texas a&m , with special focus on the neutron rich isotopes .
proton - removal and neutron pick - up isotopes have been observed with substantial cross sections . a model of deep - inelastic transfer ( dit ) for the primary interaction stage and the statistical evaporation code gemini for the deexcitation stage
have been used to describe the properties of the product distributions .
the results have also been compared with the high - energy fragmentation parametrization epax .
the experimental data show an enhancement in the production of n - rich isotopes close to the projectile relative to the predictions of dit / gemini and the expectations of epax .
we attributed this enhancement to the effect of the target @xmath1ni neutron `` skin '' in peripheral interactions of @xmath0kr with @xmath1ni .
the large cross sections of such reactions near the fermi energy , involving peripheral nucleon exchange between the projectile and the target , suggest that not only the n / z of the projectile and the target , but also the n / z distribution at the surface ( i.e. neutron `` skin '' ) may be properly exploited in the production of rare neutron - rich isotopes .
this synthesis approach may provide a fruitful pathway to extremely neutron rich nuclei towards the neutron - drip line .
we would like to thank a. sanzhur for useful discussions regarding neutron density distributions and relevant calculations .
we gratefully acknowledge the support of the operations staff of the cyclotron institute during the measurements .
financial support for this work was given , in part , by the u.s .
department of energy under grant no .
de - fg03 - 93er40773 and by the robert a. welch foundation under grant no .
a-1266 .
h. geissel and g. munzenberg , annu .
nucl . part .
45 ( 1995 ) 163 . d.j . morrissey and b. m. sherrill , phil .
lond . a 356 ( 1998 ) 1985 .
friedman , m.b .
tsang , d. bazin , and w.g .
lynch , phys .
c62 ( 2000 ) 064609 - 1 .
k. summerer and b. blank , phys .
c61 ( 2000 ) 034607 .
m. weber et al . , nucl .
a578 ( 1994 ) 659 .
l. corradi et al . , phys .
c 59 ( 1999 ) 261 .
tarasov et al . , nucl .
a629 ( 1998 ) 605 .
artukh et al . , nucl .
phys . a 176 ( 1971 ) 284 . v.v .
volkov , phys .
44 ( 1978 ) 93 .
lee et al . , phys .
c 56 ( 1997 ) 753 .
bacri et al . , nucl .
phys . a 555 ( 1993 ) 477 .
d. bazin et al . , nucl .
phys . a 515 ( 1990 ) 349 .
r. pfaff et al .
c51 ( 1995 ) 1348 .
tribble , r.h .
burch and c.a .
gagliardi , nucl .
instr . and meth . a 285 ( 1989 ) 441 .
wilcke et al . , at .
data nucl .
data tables * 25 * , 389 ( 1980 ) .
m. veselsky et al . , phys .
c62 ( 2000 ) 064613 .
r. charity , et al .
a483 ( 1988 ) 391 .
the version of gemini included modifications made up to july , 1998 .
moretto , nucl .
a247 ( 1975 ) 211 .
j. lestone , phys .
c52 ( 1995 ) 118 .
p. marmier and e. sheldon , physics of nuclei and particles , volume i ( academic , new york , 1970 ) p. 15 .
`` a proposed facility upgrade for the texas a&m univ .
cycloton institute '' , accessible at : http://cyclotron.tamu.edu/[email protected] .
cross sections and rates ( last column ) of se and ge isotopes from @xmath0kr fragmentation .
for the rates , the cross section data of this work are used and a primary beam of @xmath0kr ( 25 mev / nucleon ) of intensity 100 pna is assumed to interact with a @xmath1ni target of 20 mg/@xmath11 thickness ( see text ) . [
cols="^,^,^,^,^,^ " , ] kr + @xmath1ni .
( a ) - isobaric yield distribution .
the data are shown as solid circles ( total cross sections ) and open circles ( with acceptance cut ) .
the full line is the result of dit / gemini ( see text ) .
the dashed line is the result of the same calculation as the full line , but with a cut corresponding to the angular and momentum acceptance of the spectrometer .
( b ) - yield distributions as a function of z ( relative to the line of @xmath42 stability , z@xmath43 ) and a. highest yield contours are plotted with thicker lines .
successive contours correspond to a decrease of the yield by a factor of 2 .
the calculated values from dit / gemini are shown as i ) thick full line : without acceptance cut and , ii ) thick dashed line : with acceptance cut .
thin dashed line : epax parametrization .
( c ) - velocity vs. mass distributions .
data are shown as contours as in ( b ) .
the thick lines are as in ( b ) .
the horizontal full line represents the beam velocity .
, width=381,height=585 ] kr with @xmath1ni .
the present data are shown by full circles .
open squares are simulations according to dit / gemini and the dotted line is from the high - energy parametrization epax ( see text ) .
, width=381,height=585 ] | the cross sections and velocity distributions of projectile - like fragments from the reaction of 25 mev / nucleon @xmath0kr + @xmath1ni have been measured using the mars recoil separator at texas a&m , with special emphasis on the neutron rich isotopes .
proton - removal and neutron pick - up isotopes have been observed with large cross sections . a model of deep - inelastic transfer ( dit ) for the primary interaction stage and the statistical evaporation code gemini for the deexcitation stage
have been used to describe the properties of the product distributions .
the results have also been compared with the epax parametrization of high - energy fragmentation yields .
the experimental data show an enhancement in the production of neutron - rich isotopes close to the projectile , relative to the predictions of dit / gemini and the expectations of epax .
we attribute this enhancement mainly to the effect of the extended neutron distribution ( neutron `` skin '' ) of the @xmath1ni target in peripheral interactions of @xmath0kr with @xmath1ni .
the large cross sections of such reactions near the fermi energy , involving peripheral nucleon exchange , suggest that , not only the n / z of the projectile and the target , but also the n / z distribution at the nuclear surface may properly be exploited in the production of neutron - rich rare isotopes . this synthesis approach may offer a fruitful pathway to extremely neutron - rich nuclei , towards the neutron - drip line .
@xmath2 , , , , , , , , and . rare isotope production , nuclear reactions , deep inelastic transfer , fermi energy , neutron `` skin '' 25.70.-z , 25.70.hi , 25.70.lm |
a tonks - girardeau gas @xcite is a one - dimensional ( 1d ) system of identical bosons with strong interparticle interactions .
the wave functions of the bosons have a surprisingly simple one - to - one mapping to a system of non - interacting fermions @xcite . in recent years
, a tonks - girardeau gas was realized in experiments with ultracold gases @xcite .
a characteristic property of the tonks - girardeau gas is that the probability to find two particles at the same position is very small @xcite .
theoretical and experimental studies of the tonks - girardeau gas so far dealt only with the case of elastic interactions . in a recent experiment , we studied a 1d gas in which the bosons strongly interact inelastically , leading to loss of particles @xcite .
this system also shows a strong suppression of the probability to find two particles at the same position and , indeed , demonstrates an alternative way of realizing a tonks - girardeau gas .
the broader impact of this experiment lies in the demonstration that dissipation can be used to drive a system into the strongly - correlated regime . here
we study the theory of the dissipative tonks - girardeau gas .
we show that in the limit of infinite interaction strength girardeau s solution @xcite is reached , as in the case of strong elastic interactions . we derive a markovian master equation that describes the dissipation in the system and yields a rate equation for the particle loss .
when the study is restricted to loss rates at short times , the master equation can be reduced to a schrdinger equation with a non - hermitian version of the lieb - liniger hamiltonian @xcite .
this approach yields an analytic expression for the loss rate in the limit of strong dissipation .
furthermore , we present numerical solutions to the dissipative lieb - liniger model at finite interaction strength that illustrate how a possible elastic part of the interactions becomes irrelevant as the dissipation strength diverges . in addition
these numerical calculations reveal the dependence of the pair correlation function on dissipation strength , showing that an increase in the dissipation strength leads to a decrease of the actual loss rate due to the build - up of strong correlations
. finally , we present an estimate for the loss rate at longer times .
in this section , we discuss a master equation approach to describe loss processes caused by inelastic two - body collisions in an ultracold gas .
first , we introduce the master equation , then use it to derive a rate equation for the particle loss , and finally show that in the short - time limit , a non - hermitian lieb - liniger model is obtained .
we consider a three - dimensional ( 3d ) , dilute , ultracold gas of identical bosons , which all occupy the same internal state .
two - body collisions in such a gas can be described in terms of the 3d @xmath0-wave scattering length @xmath1 . for elastic collisions ,
it is customary to replace the interparticle scattering potential by a delta potential @xmath2 with interaction strength @xcite @xmath3 where @xmath4 is the mass of a particle . for the many - body system in second quantization
, this yields a hamiltonian ( see appendix [ app - master ] or , _
e.g. _ , ref .
@xcite ) @xmath5 where @xmath6 is the single - particle hamiltonian in first quantization and @xmath7 is the field operator that annihilates a boson at position @xmath8 . in the case of purely elastic interactions
@xmath9 is real anyway , so taking the real part of it in eq .
( [ h-0 ] ) has no effect .
we now generalize this treatment to include inelastic collisions .
first , we consider inelastic collisions on the level of a two - body problem .
this yields an imaginary part of the scattering length , with @xmath10 @xcite .
this is because the elastically scattered wave is proportional to the diagonal element of the @xmath11-matrix @xmath12 , where @xmath13 is the wave vector in the relative motion .
hence , a negative imaginary part of @xmath1 reduces the elastically scattered flux .
this missing flux appears in the inelastically scattered channels .
next , we consider the effect of inelastic collisions on the level of the many - body system .
the interaction strength @xmath9 defined by eq .
( [ g-3d ] ) becomes complex - valued .
equation ( [ h-0 ] ) contains only the real part of @xmath9 and thus represents only the elastic part of the interactions ( see appendix [ app - master ] ) . for a treatment of the inelastic collisions , it is crucial what happens to the collision partners after the collision .
the difference in internal energy that is released in the change of the internal state appears in the kinetic energy of the relative motion of the particles after the collision . in typical experiments with trapped ultracold gases ,
the energy released here is so large that all particles involved in the collision quickly escape from the trap .
this loss of particles is an irreversible process . in appendix
[ app - master ] , we derive a markovian quantum master equation @xmath14 + { \cal d}(\rho)\end{aligned}\ ] ] for the time evolution of the density matrix @xmath15 of this many - body system , with a dissipator @xmath16 that has a lindblad form @xcite .
the dissipator describes loss of pairs of particles due to inelastic collisions .
more specifically , the last two terms in @xmath17 deplete terms in the density matrix that represent states with more than one particle .
the first term in @xmath17 makes this lost population reappear in states with two fewer particles .
this is further illustrated in sec .
[ sec - short - times ] .
we now use this master equation approach to derive a rate equation for the loss of particle number .
the particle density operator @xmath18 has an expectation value that has a time dependence with a contribution from @xmath19 , which conserves the total particle number , and a contribution from @xmath20 , which describes loss @xmath21 \\ \label{3d - loss - temp } & = & -k_{\rm 3d } \langle \psi^{\dagger 2}({\bf x } ) \psi^2({\bf x})\rangle , \end{aligned}\ ] ] where tr denotes the trace and @xmath22 is the rate coefficient for two - body loss . the loss - rate equation ( [ 3d - loss - temp ] ) can be rewritten as @xmath23 with the pair correlation function @xmath24 if @xmath25 particles occupy the same single - particle wave function , such as in a bose - einstein condensate ( bec ) , one obtains @xmath26 . for @xmath27
such a system is uncorrelated , _
i.e. _ , @xmath28 . for comparison ,
a thermal gas of bosons above the critical temperature for bose - einstein condensation shows thermal bunching with @xmath29 .
@xmath30 quantifies how far the loss rate deviates from that of an uncorrelated system .
similar considerations apply to inelastic three - body collisions @xcite .
the above master - equation approach , has a straightforward generalization to a 1d system , yielding @xmath31 where @xmath32 is the 1d particle density and @xmath33 is the strength of the 1d delta potential , which can be related to the 3d scattering length @xmath1 as described now .
experimental realizations of 1d systems typically use a strong harmonic confinement in the two transverse dimensions , with trap angular frequency @xmath34 .
if one approximates the 3d particle density along the two transverse dimensions by the ground state of the harmonic oscillator with oscillator length @xmath35 , then spatial integration of eq .
( [ dndt-3d ] ) over the two transverse dimensions yields the estimate @xmath36 a more rigorous approach to this scattering problem in a confined geometry was presented in ref .
@xcite , yielding @xmath37^{-1 } , \end{aligned}\ ] ] where @xmath38 denotes the riemann zeta function with @xmath39 . in the limit @xmath40 , this reproduces eq .
( [ k1d - approx ] ) .
but outside this regime , the scattering process is altered by the transverse confinement and eq .
( [ k1d - approx ] ) is no longer valid .
the derivation in ref .
@xcite considered only real - valued @xmath1 , but following this derivation , one can show that eq .
( [ g-1d ] ) remains valid for complex - valued @xmath1 .
the applicability of eq .
( [ g-1d ] ) for the parameters of our experiment in ref .
@xcite is discussed in appendix [ app - experiment ] . for studies of the behavior of the system at short times ,
it is useful to rewrite the master equation ( [ master ] ) as @xmath41 with an effective hamiltonian , that is not hermitian and turns out to be the analytic continuation of @xmath19 @xmath42 and with @xmath43 note that in a quantum monte - carlo approach to open quantum systems , @xmath44 would be represented by the quantum jump part of the time evolution @xcite .
the effect of the different terms in the master equation ( [ master - eff ] ) can be illustrated by considering a density matrix that initially represents a homogeneous bec with exactly @xmath25 particles .
for short times eq .
( [ master - eff ] ) yields @xmath45 where @xmath46 .
the loss rate of particle number is @xmath47 .\end{aligned}\ ] ] a detailed look at the calculation shows that @xmath48 causes the decay of the probability to find @xmath25 particles , whereas @xmath44 causes the build - up of the probability to find @xmath49 particles .
if we are interested only in calculating @xmath50 at short times , @xmath51 , we can use this observation to drastically simplify the model .
the key idea is that refilling population into states with @xmath49 particles is important only when the evolution at long times is considered , namely at times where states with @xmath49 particles also decay significantly .
but at short times , all that counts is how fast population is lost from the initial state . guided by this idea
, we drastically simplify the model by dropping @xmath44 from the model .
this removes the term @xmath52 from eq .
( [ xi ] ) . as a result
, @xmath53 decays as a function of time , which is unphysical , but the rate at which this decay occurs @xmath54\end{aligned}\ ] ] is quite informative , because this is the rate at which population is lost from the initial state .
we know that in reality this population reappears in states with @xmath49 particles .
as each loss removes two particles , the true loss rate of particle number can be estimated from the decay rate of @xmath53 in the simplified model . indeed ,
comparison of eqs .
( [ dndt - xi ] ) and ( [ dtrrhodt - xi ] ) yields @xmath55 note that here @xmath50 refers to the true loss rate of particle number in the full model , whereas @xmath56 is a quantity calculated in the simplified model .
the factor of 2 in this equation reflects the fact that two particles are lost per inelastic collision . with @xmath15 initially prepared in a pure state @xmath57 and with @xmath44 neglected
, the master equation reduces to a schrdinger equation with a non - hermitian hamiltonian @xmath48 . according to eq .
( [ dndt - tr - rho ] ) the particle loss rate becomes @xmath58 for a right eigenvector of the effective hamiltonian @xmath59 , we obtain @xmath60 the calculation of the initial loss rate is thus possible using a schrdinger equation with @xmath48 , instead of the master equation .
the imaginary part of an energy eigenvalue @xmath61 is always non - positive and represents the initial loss rate of the particle number .
this greatly simplifies the calculation , because it suffices to study the eigenvectors and eigenvalues of an effective hamiltonian , instead of performing a time - resolved calculation of the density matrix .
we will follow this approach throughout sec .
[ sec - lieb ] .
motivated by the results of the previous section , we now study the eigenvectors and eigenvalues of the effective hamiltonian .
this effective hamiltonian is a non - hermitian version of the lieb - liniger model @xcite .
we are mostly interested in the regime of strong dissipation .
first , we show that in this regime , a tonks - girardeau gas is reached , second , we derive an analytical expression for the pair - correlation function in this regime and , third , we compare this with results of a numerical calculation , which show that a finite real part of the scattering length becomes irrelevant for diverging dissipation strength .
we consider a 1d gas of @xmath25 identical bosons in a box of length @xmath62 with periodic boundary conditions and with the 1d version of the effective hamiltonian of eq .
( [ h - eff ] ) . in first quantization
this effective hamiltonian reads @xmath63 where @xmath64 is the position of the @xmath65th boson . for the case of real - valued @xmath1 ,
this model was first considered by lieb and liniger @xcite , who introduced the dimensionless interaction strength @xmath66 we explain now why this model in the limit @xmath67 yields exactly girardeau s solution @xcite , irrespective of whether the interactions are elastic or inelastic .
the delta potential in eq .
( [ h - ll ] ) can be replaced by the following boundary condition at positions where the relative coordinate @xmath68 vanishes @xcite @xmath69 bosonic symmetry implies that on the left - hand side , the first term equals minus the second term .
expanding @xmath70 in a power series for @xmath71 , eq .
( [ boundary ] ) can be rewritten as @xmath72 + |x_{ij}| + { \cal o}(x_{ij}^2 ) , \end{aligned}\ ] ] where the proportionality contains the dependence on all remaining coordinates .
the energy eigenstates of this model can be divided into two classes : gaseous states and bound states . as an example of a bound state
, we consider a state where @xmath73 and where particles @xmath65 and @xmath74 are bound . here
@xmath75 for @xmath76 , which obviously meets the condition ( [ boundary - ver2 ] ) .
an expansion in a basis of plane waves @xmath77 shows that this state contains imaginary momentum components that diverge for @xmath78 . bound states exist only if @xmath73 .
the case of real and negative @xmath79 was studied , _
e.g. _ , in refs .
@xcite . for real @xmath79
, two characteristic properties can be used to distinguish between bound and gaseous states : first , the momenta of all particles are real valued for gaseous states , whereas at least one momentum is complex for bound states .
second , for @xmath80 all momenta converge to finite values for gaseous states , whereas at least one momentum diverges to @xmath81 for bound states .
we consider the case of complex @xmath79 , where all momenta are usually complex so that the first characteristic property becomes useless , but the second one remains useful , namely the convergence or divergence of the momenta for @xmath67 can still be used to distinguish between gaseous and bound states .
this distinction implies that @xmath70 and @xmath82 remain finite for @xmath83 for all gaseous states .
hence , the left - hand side of eq .
( [ boundary ] ) remains finite for @xmath83 , so that the right - hand side must remain finite , too . with @xmath33 diverging
, the right - hand side implies that @xmath84 for all gaseous states .
the same result can be obtained from eq .
( [ boundary - ver2 ] ) , because the term proportional to @xmath85 must remain finite so that the proportionality factor must remain finite . with @xmath33 diverging , the constant term vanishes .
yet another way to obtain eq .
( [ psi - vanishes ] ) is to consider a right eigenvector of the effective hamiltonian @xmath86 .
if @xmath87 for at least one relative coordinate @xmath88 , then eq . ( [ h - ll ] ) shows that @xmath89 , so that the state decays with infinite speed .
we decompose @xmath90 and see that after any nonzero time only those eigenstates with finite @xmath91 survive . again , this yields eq . ( [ psi - vanishes ] ) .
the other states are bound states .
they vanish immediately and no longer contribute to the state or to the dynamics of the system . in the limit @xmath67 ,
the interaction for all gaseous states is fully described by the boundary condition ( [ psi - vanishes ] ) .
the crucial point is that this boundary condition is independent of whether @xmath79 is real or complex , and it is precisely this boundary condition that yields a tonks - girardeau gas .
hence , the wave functions of all gaseous states turn exactly into girardeau s solutions @xcite in the limit @xmath67 , irrespective of whether @xmath79 is real or complex . attraction , repulsion , and dissipation all produce the tonks - girardeau gas in the limit of infinite interaction strength .
the above discussion shows that an initially uncorrelated state subject to strong dissipation will experience a rapid initial decay .
some fraction of the population will survive and this remaining population will be in the tonks - girardeau subspace , where the loss is slow
. the loss will prevent the system from subsequently moving far away from the tonks - girardeau subspace .
this freezing of the population can be understood intuitively with an analogy from classical optics @xcite or as a manifestation of the quantum zeno effect @xcite .
the focus of our study is the behavior of the system once it has reached the tonks - girardeau subspace , not the rapid initial decay .
the experiment in ref .
@xcite started from a strongly correlated state already , so that no rapid initial decay occurred . here , we summarize the central results of lieb and liniger @xcite insofar as they are required for the following discussion .
derivations for all results of this section can be found in ref .
following these derivations , one can show that these results also apply to the case of complex @xmath33 .
the lieb - liniger solution is based on a bethe ansatz @xcite @xmath92 which holds only for @xmath93 . for other values of the coordinates
, the solution is obtained from bosonic symmetry and periodic boundary conditions .
the sum extends over all permutations @xmath94 of the numbers @xmath95 .
the parameters @xmath96 need to be determined . once they are known , the amplitudes @xmath97 can be calculated easily @xcite .
the bethe ansatz yields a nonzero solution for @xmath70 only if all @xmath98 are mutually different @xcite .
the energy eigenvalue of this solution is @xmath99 lieb and liniger characterize the solutions by quantum numbers @xmath100 , which are positive integers . for a given set of these quantum numbers the following set of coupled implicit equations
is to be solved for @xmath101 @xcite @xmath102 where @xmath103 abbreviates @xcite @xmath104 with @xmath105 .
this set of @xmath106 implicit equations determines only the @xmath106 quantities @xmath107 .
there is one remaining degree of freedom .
it can be expressed in terms of the quantum number @xcite @xmath108 which must also be an integer . for a given set of quantum numbers @xmath100 ,
the value of @xmath109 is constrained by the additional condition @xcite @xmath110 if one restricts @xmath109 to the range @xcite @xmath111 then there is exactly one value @xmath109 for each combination of the @xmath100 . adding @xmath25 to @xmath109 adds a momentum of @xmath112 to each particle in the system .
this corresponds to a center - of - mass motion of the gas as a whole , which is trivial and of little interest .
the values of @xmath109 in the range ( [ n0-range ] ) , however , correspond to different internal quantum numbers and thus need to be taken into account @xcite .
while the bethe ansatz does yield solutions for complex @xmath79 , it remains unclear whether a complete set of solutions is obtained .
our numerical results , discussed further below , suggest that each solution for real ( positive or negative ) @xmath79 continuously deforms into exactly one solution for complex @xmath79 , thus suggesting that completeness for real @xmath79 and complex @xmath79 are closely linked . but
a solid proof of completeness for complex @xmath79 is beyond the scope of the present paper .
on the other hand , the completeness issue has only limited relevance for the experiment discussed in ref .
@xcite , because here only the time evolution of specifically - prepared gaseous initial states is to be tracked .
we now consider the limit @xmath113 .
we expand the @xmath114 in eq .
( [ theta - ij - arctan ] ) in a power series and keep only the lowest order .
this requires that @xmath115 vanishes for large @xmath116 .
a sufficient condition is that all @xmath98 remain finite , which defines the gaseous states .
for any large but finite @xmath116 , the gaseous states can be divided into two classes : first , the so - called fermionized states for which @xmath117 so that we can expand the @xmath114 and , second , the other states which must have very large @xmath98 so that they are highly excited . in the limit @xmath113 all gaseous states become fermionized . insertion of this expansion into eq .
( [ n - j ] ) yields @xmath118 iterating this equation , we obtain for @xmath119 @xmath120 insertion into eq .
( [ n-0 ] ) yields @xmath121 insertion into eq .
( [ periodic - boundary ] ) yields that @xmath122 must be an integer , thus fixing @xmath109 modulo @xmath25 .
as @xmath109 must generally be an integer , the term @xmath123 must vanish automatically . for given values of @xmath79 and @xmath100
, we thus obtain @xmath109 and all the @xmath98 .
these general results for fermionized states can be used to calculate the ground state properties .
the ground state of the gas phase is characterized by @xcite @xmath124 eqs .
( [ k - j ] ) - ( [ get - n-0 ] ) yield @xmath125 and @xmath126 according to eq .
( [ ll - energy ] ) the ground state energy is @xmath127 this result was previously derived by lieb and liniger @xcite in an alternative way for the case of real @xmath79 and in the limit @xmath128 .
our above calculation generalizes this result to complex @xmath79 and finite @xmath25 .
expansion to lowest order in @xmath129 , taking the limit @xmath27 , and insertion into eq .
( [ dn - dt - im - e ] ) yield the particle loss rate at short times @xmath130 we use @xmath131 and compare with eqs .
( [ dndt-1d ] ) and ( [ k-1d ] ) .
this yields one of the central results of this paper @xmath132 for the special case of real @xmath79 , this result was previously obtained in ref .
note that @xmath133 for @xmath83 , irrespective of whether @xmath79 is real or complex .
this is a consequence of the boundary condition ( [ psi - vanishes ] ) .
real and imaginary parts of @xmath134 for the @xmath135 ground state with @xmath136 .
the same data are plotted in a different way as the top - left curve in fig .
[ fig - complex - k - plane](a ) . ]
( [ dndt - ground - state ] ) reveals that the loss rate depends on the scattering properties only in terms of @xmath137 .
we can rewrite eq .
( [ g-1d ] ) as @xmath138 and find that @xmath139 , irrespective of whether the condition @xmath40 is met or not . a possible alteration of the scattering process in the regime where the condition @xmath40 is not met affects @xmath33 in such a way that the loss rate in the fermionized regime is unchanged . in the fermionized regime ,
the dependence of the loss rate on the scattering length is always given by @xmath140 note that in a case where an additional lattice is applied along the one dimension , the loss rate also obeys eq .
( [ dndt - ima ] ) , see appendix [ app - experiment ] and refs .
@xcite .
solutions of the lieb - liniger model outside the limit @xmath67 can be found numerically . to this end
, we numerically solve the set of coupled eqs .
( [ n - j ] ) and ( [ theta - ij - arctan ] ) with the standard root - finding algorithm in _
mathematica_. computation time issues restrict this approach to reasonably small @xmath25 .
it is not necessary to choose a specific value of @xmath62 for the numerical calculations , because the relevant equations can be written in terms of the dimensionless parameters @xmath25 , @xmath79 , and @xmath141 .
we illustrate these solutions for the ground state characterized by eq .
( [ all - n=1 ] ) with @xmath135 particles .
figure [ fig - k - vs - gamma ] shows the real and imaginary parts of one of the parameters , namely @xmath134 , as a function of @xmath142
. an alternative way of displaying the same data is to show the trajectory that @xmath134 follows in the complex @xmath143 plane when @xmath144 is scanned from 0 to @xmath145 .
such plots are shown in fig .
[ fig - complex - k - plane ] for all @xmath98 .
parts ( a ) , ( b ) , and ( c ) each correspond to one fixed value of @xmath146 .
for @xmath147 the solutions in all parts of the figure converge to girardeau s solution ( shown as filled circles ) .
the important result is that @xmath146 becomes irrelevant for @xmath148 and that girardeau s solution is reached in any case .
the values of @xmath98 obtained in such numerical calculations can be used to extract @xmath30 in the same way as in the derivation of eq .
( [ g2-ground - state ] ) .
results for the ground state are shown in fig .
[ fig - g2](a ) .
@xmath30 depends on @xmath25 , but converges in the thermodynamic limit . the solid line in fig .
[ fig - g2](a ) shows the case @xmath149 . on the scale shown here it
can not be distinguished from the case @xmath150 ( not shown ) .
this suggests that fig .
[ fig - g2](a ) shows a reasonable approximation to the thermodynamic limit .
a thorough analysis of finite - size effects is beyond the scope of this work .
parametric plot , showing the @xmath98 in the complex @xmath143 plane for the @xmath135 ground state .
each curve shows a trajectory traced by @xmath98 when scanning the parameter @xmath144 from 0 to @xmath145 . in all cases ,
the curves begin for @xmath151 somewhere not too far from the coordinate origin and converge for @xmath152 to girardeau s solution ( shown as filled circles ) .
( a ) @xmath136 . for @xmath153 all @xmath154 .
( b ) @xmath155 . for @xmath151 ,
the @xmath98 are approximately equally spaced on the real axis .
( c ) @xmath156 . for @xmath151 ,
the @xmath98 are approximately equally spaced on the imaginary axis , indicating a bound state with an energy that is real and negative . however , as @xmath157 grows and becomes larger than @xmath158 , the finite @xmath146 becomes irrelevant and girardeau s solution is reached for @xmath67 . ]
the loss rate eq .
( [ dndt-1d ] ) can be rewritten as @xmath159 the dependence on @xmath79 is given by the dimensionless expression @xmath160 .
this quantity is displayed in fig .
[ fig - g2](b ) , showing that for @xmath161 the loss suppresses itself .
the experiment in ref .
@xcite was performed for @xmath162 which is in this regime . pair correlation function and loss rate of the ground state as a function of the dissipation strength for @xmath136 and @xmath149 .
( a ) the pair correlation function @xmath30 ( solid line ) equals @xmath163 at @xmath153 and vanishes for @xmath164 .
for @xmath165 it is well approximated by eq .
( [ g2-ground - state ] ) ( dotted line ) .
( b ) the dimensionless quantity @xmath166 is proportional to the particle loss rate , see eq .
( [ dndt-1d - gamma ] ) . for small @xmath116 the loss rate increases linearly with @xmath144 because @xmath30 is approximately constant .
but for large @xmath116 , the suppression of @xmath167 dominates and the loss rate vanishes like @xmath168 .
this is remarkable : the loss suppresses itself . ]
so far we studied only the loss rate at short times .
a detailed study of loss rates at longer times would require a numerical solution of the master equation including the term @xmath169 .
this is beyond the scope of the present paper , but for the case with a lattice applied along the one dimension , results of such a numerical calculation are reported in ref .
@xcite . instead of solving the master equation
, we now present a fairly simple estimate for the temporal evolution of the loss rate . during the loss
, a spread in particle number builds up , but this spread is expected to reach a poisson distribution , which is narrow enough that it suffices to consider only the mean value of the particle number . as the loss proceeds , the particle number changes and so does the ground - state energy @xmath170 in eq .
( [ e - ground - state ] ) . in the lieb - liniger model with periodic boundary conditions , @xmath171 . in a more realistic model with a harmonic trap along the one dimension ,
the scaling is @xmath172 , which we use in the following .
a simple estimate can be obtained when assuming that the average kinetic ( _ i.e. _ , real ) part of the energy per particle is preserved by the loss , so that @xmath173 remains constant .
we assume that @xmath174 at @xmath175 .
as @xmath25 decreases , @xmath176 will then increase according to @xmath177/{\rm re}[e_g(t)]=n(0)/n(t)$ ] .
the experiment in ref .
@xcite monitored a decrease of @xmath25 by a factor of @xmath178 .
our simple estimate predicts an increase of @xmath177/{\rm re}[e_g(t)]$ ] by the same factor of 2 .
we conclude that while the loss occurs , the energy of the system evolves somewhat away from the energy of the ground state , but not terribly far .
this simple estimate is based on the _ ad hoc _ assumption that @xmath173 is preserved by the loss .
we now study how realistic this assumption is . as the loss takes the system into the regime @xmath177>{\rm re}[e_g(t)]$ ]
, we need to study the excited states of the gas phase .
we restrict our considerations to the fermionized regime . in a first step , we ignore the condition ( [ periodic - boundary ] ) and set @xmath125 .
( [ ll - energy ] ) , ( [ k - j ] ) , and ( [ k-1 ] ) yield @xmath179 where @xmath180 as the ground state
truly has @xmath125 we obtain @xmath181 where @xmath170 denotes the ground - state energy .
obviously @xmath182 is always real .
hence , @xmath183 we will show now that the correction for @xmath184 is negligible . to this end , we note that according to eqs .
( [ k - j ] ) and ( [ k-1 ] ) , @xmath109 shifts all @xmath98 by @xmath185 . as a result ,
the energy becomes @xmath186 eq .
( [ n0-range ] ) yields @xmath187 . in the thermodynamic limit ( @xmath128 at fixed @xmath32 ) , the correction is of order @xmath188 . on the other hand , @xmath189 .
hence , @xmath190 therefore the correction due to @xmath109 is negligible and eq .
( [ e - n=0-e0 ] ) becomes @xmath191 this shows that the ratio of the real to the imaginary part of @xmath61 is identical for all fermionized states , independent of @xmath100 .
combined with eq .
( [ dn - dt - im - e ] ) we obtain that the loss rate of a state is proportional to the kinetic ( _ i.e. _ , real ) part of the energy of a state .
this has two important consequences for the understanding of the loss at longer times .
first , it shows that an increase of @xmath177/{\rm re}[e_g(t)]$ ] by a given factor leads to an increase of the loss rate compared to that of the ground state by the same factor .
second , it shows that the loss preferentially depletes states with large energy .
hence , the above estimate assuming that the loss leaves the average kinetic energy per particle unchanged is too pessimistic . in reality , the ratio @xmath177/{\rm re}[e_g(t)]$ ] increases even more slowly than predicted by the simple estimate above .
the kinetic energy and loss rate of the system will thus remain fairly close to those of the ground state if the loss is allowed to evolve only for a time where the particle number decays by a factor of @xmath178 as in ref .
in conclusion , we showed that strong inelastic interactions produce a tonks - girardeau gas , much as strong elastic interactions .
we derived an analytic expression for the pair correlation function of the fermionized ground state .
numerical results illustrate the behavior of the ground state for small and medium dissipation strength .
these calculations also show that the loss suppresses itself .
finally , we presented a simple estimate for the evolution of the loss rate for longer times .
in this appendix we derive the master equation ( [ master ] ) with the dissipator ( [ d ] ) for the description of loss of particles from a many - body system due to ultracold inelastic two - body collisions .
the system of interest @xmath11 consists of trapped motional states in the initial internal state .
inelastic collisions couple this system @xmath11 to other states , which we call the reservoir @xmath192 .
the reservoir states can represent all possible internal states .
the motional parts of the reservoir states are untrapped because we assumed that all particles involved in an inelastic collision quickly escape from the trap .
hence the energy spectrum of these untrapped states is a continuum .
the total system consisting of @xmath11 and @xmath192 is represented by a density matrix @xmath193 .
as we are not interested in the reservoir , we can trace out the reservoir degrees of freedom and obtain a density matrix @xmath15 that describes only the system of interest @xmath11 .
we assume that the reservoir is in the vacuum state initially and we use a born - markov approximation @xcite for the reservoir modes , _ i.e. _ , we assume that the reservoir has no memory of the past .
this is justified because the reservoir consists of a continuum of modes that quickly dephase . as a result
, the reservoir returns to the vacuum state on a time scale , which is much faster than all time scales in the system of interest @xmath11 .
using very general arguments one can show @xcite that such a born - markov approximation yields a markovian quantum master equation @xmath194 + { \cal d}(\rho)\end{aligned}\ ] ] for the evolution of the density matrix @xmath15 describing only the system of interest @xmath11 .
this master equation contains a hamiltonian @xmath195 and a so - called dissipator @xmath17 , which models the irreversible part of the time evolution and can always be written in a lindblad form @xcite @xmath196 where the @xmath197 are relaxation rate coefficients and the @xmath198 are called lindblad operators .
the rest of this appendix is devoted to deriving expressions for @xmath195 and @xmath17 for our particular system .
given the form of the elastic collision term in eq .
( [ h-0 ] ) , the final result for the lindblad operator @xmath199 in eq .
( [ d ] ) is not terribly surprising .
but the result for the rate coefficient @xmath200 is certainly less obvious , in particular as far as possible factors of 2 are concerned .
we now present a rigorous derivation for this dissipator , including the rate coefficient .
the calculation proceeds in three steps .
first , we determine a contact potential that can be used to describe inelastic scattering , second , we derive the corresponding hamiltonian in second quantization and , third , we eliminate the reservoir degrees of freedom . for the first step
, we consider a scattering process with two distinguishable incoming particles and assume , for simplicity , that after the collision , there are two outgoing particles with the same mass as the incoming particles .
we further assume that the incoming wave is an @xmath0 wave and that the potential is spherically symmetric , so that the outgoing wave is also an @xmath0 wave .
the scattering state @xmath201 at large relative distance of the particles is @xmath202 where @xmath203 is the relative coordinate , @xmath204 is a matrix element of the scattering matrix , and the internal states of the particles before and after the collision are @xmath205 and @xmath206 with corresponding wave vectors for the relative motion @xmath207 and @xmath208 , respectively .
the wave function ( [ r - psi ] ) can be regarded as a generalization of eq .
( 5 ) in ref . @xcite to inelastic @xmath0-wave scattering .
the fact that the outgoing particle flux is proportional to @xmath209 makes it necessary to include the factor @xmath210 in eq .
( [ r - psi ] ) , in order to maintain the fact that the @xmath11-matrix is unitary if the number of particles is conserved in the collision .
we now replace the interparticle interaction potential by a contact potential .
the wave function ( [ r - psi ] ) that is usually valid for @xmath211 only , then becomes valid for all @xmath212 .
we choose a contact potential @xmath213 with a part @xmath214 that acts only on the internal state and a part @xmath215 that acts only on the spatial part of the wave function .
for the latter , we use the regularized delta function @xmath216 ( see , _ e.g. _ , refs .
in addition , we assume that @xmath214 is hermitian .
we expand the wave function ( [ r - psi ] ) in a power series for @xmath217 and spatially integrate the schrdinger equation @xmath218 over a sphere centered around @xmath219 with radius @xmath220 .
we use @xmath221 and @xmath222 .
we neglect terms @xmath223 .
this yields for all @xmath224 @xmath225 with the reduced mass @xmath226 and the kronecker symbol @xmath227 . we do not attempt to solve this linear system for the variables @xmath228 in full generality .
instead , we decide to fulfill eq .
( [ coeff - zero ] ) for only one specific initial state @xmath229 .
we thus assume @xmath230 this yields @xmath231 where we used that unitarity of the @xmath11-matrix implies @xmath232 .
the scattering phase @xmath233 , defined by @xmath234 , can be used to rewrite the above results in the compact form @xmath235 for ultracold collisions , the ratio @xmath236 becomes constant and is called the scattering length @xmath237 the above treatment applies to scattering of distinguishable particles , while what we are really interested in is scattering of identical bosons . due to the linearity of the schrdinger equation , scattering of identical bosons
is closely related the scattering of distinguishable particles from the same potential .
our following approach is based on second quantization , which has bosonic symmetry built into it automatically , so that no separate treatment of scattering of identical bosons is needed here .
the interactions described by the above contact potential can be translated to a hamiltonian in second quantization for a many - body system . if a general two - body interaction potential @xmath238 is summed over all possible pairs of particles , one obtains @xmath239 , where @xmath240 enumerate the particles . in second quantization , this yields a hamiltonian @xcite @xmath241 where the states @xmath242 form a basis of single - particle states and @xmath243 creates a boson in state @xmath242 .
we now apply this to our contact potential . as a basis of single particle states ,
we choose @xmath244 , where @xmath245 is an internal state and @xmath246 a momentum state with @xmath247 in box quantization with quantization volume @xmath248 .
we denote the corresponding creation operator as @xmath249 .
a short calculation yields the scattering hamiltonian for our contact potential @xmath250 where @xmath251 is the center - of - mass momentum of a pair of bosons , and @xmath252 and @xmath253 are the relative momenta before and after the collision .
we denote the corresponding kinetic energies as @xmath254 , @xmath255 , and @xmath256 .
note that in eq .
( [ h - sc ] ) it makes no difference if we take @xmath257 or @xmath258 .
let us assume for a moment that the @xmath259 are independent of @xmath260 , @xmath261 , and @xmath262 .
@xmath263 can then be transformed into a simple expression in position representation @xmath264 where the field operator @xmath265 creates a boson at position @xmath266 in internal state @xmath245 .
note that the term in @xmath263 that describes elastic scattering in the initial internal state is proportional to @xmath267 from eq .
( [ g - ii ] ) . using eq .
( [ a ] ) , we obtain exactly the elastic collision term in eq .
( [ h-0 ] ) .
the resulting mean - field energy is small compared to the kinetic energies involved in the following calculation .
hence we can perform an approximation of independent rates of variation @xcite .
this means that we include the elastic collision term in the final eq .
( [ h-0 ] ) , but we neglect its effect on @xmath17 by setting @xmath268 in the following calculation . having determined the explicit form of @xmath269 , we now proceed to eliminate the reservoir modes . to this end , we split the total hamiltonian of our problem into three parts : @xmath270 acting only on the system of interests @xmath11 , @xmath271 acting only on the reservoir @xmath192 , and @xmath272 coupling @xmath11 and @xmath192 .
the single - particle hamiltonian is thus decomposed into @xmath273 where the term @xmath274 belongs to @xmath270 , all others to @xmath271 .
here @xmath275 is the internal energy in internal state @xmath245 .
we choose the zero of internal energies such that @xmath276 . for simplicity , we neglect collisions except if they make a transition between @xmath11 and @xmath192 .
hence @xmath277 with the @xmath259 from eqs . ( [ g - zero])([g - hermitian ] ) and with the assumption @xmath268 .
we further simplify the model by excluding the possibility that a reservoir state contains any particles in the initial internal state @xmath278 .
this is a simple way to reflect the fact that all particles involved in an inelastic collision are lost from the sample .
our discussion throughout the rest of this section closely follows chapter 1 of ref .
the total system consisting of @xmath11 and @xmath192 is described by a density matrix @xmath279 that obeys the von - neumann equation @xmath280 $ ] .
we transform this into an interaction picture @xmath281 \label{von - neumann } .\end{aligned}\ ] ] note that @xmath282 picks up an explicit time dependence .
we formally integrate the last equation @xmath283 \end{aligned}\ ] ] and substitute this result into the commutator in eq .
( [ von - neumann ] ) @xmath284 \nonumber \\ & & - \frac{1}{\hbar^2 } \int_0^t dt ' [ \tilde h_{sr}(t),[\tilde h_{sr}(t'),\tilde \chi(t ' ) ] ] .\end{aligned}\ ] ] we assume that @xmath192 is in the vacuum state at @xmath175 , uncorrelated with the state of @xmath11 .
we assume that @xmath272 is weak and that the reservoir @xmath192 is much larger than the system @xmath11 , so that the state of @xmath192 is hardly affected by the interaction with @xmath11 .
the correlations between @xmath11 and @xmath192 that build up during the time evolution are then weak . in a born approximation @xcite ,
we neglect these correlations and write @xmath285 , where @xmath286 is the vacuum density matrix for @xmath192 .
we insert this into the right - hand side of the last equation .
we then take the partial trace @xmath287 over the reservoir , assume @xmath288=0 $ ] and obtain @xcite @xmath289\ } .
\quad\end{aligned}\ ] ] we write the scattering hamiltonian ( [ h - sc ] ) with the @xmath259 from eqs .
( [ g - zero])([g - hermitian ] ) and with the assumption @xmath268 as @xmath290 we recall our assumption that a reservoir state can not have any particles in state @xmath278 .
hence , @xmath291 acts only on @xmath11 ( internal state @xmath278 ) , whereas @xmath292 acts only on @xmath192 ( all other internal states ) .
in addition , our above assumption @xmath288=0 $ ] is satisfied .
moreover , @xmath293 because swapping which internal state is labeled first , can not change the physics .
hence @xmath294 .
insertion of @xmath272 into eq .
( [ rho - tilde - dot ] ) shows that @xmath287 vanishes for most of the terms .
we obtain @xmath295 r_{\bf k}(t , t ' ) + h.c .
, \quad\end{aligned}\ ] ] where we abbreviated @xmath296 .
we use @xmath297 and obtain @xmath298 for @xmath299 or @xmath300 ( or both ) the last line equals @xmath301 using @xmath294 we see that both terms produce identical results after carrying out @xmath302 . for @xmath303 and @xmath304 , we obtain only one term , but after carrying out @xmath302 , the result is the same because @xmath305 .
hence , @xmath306 we now employ a continuum approximation for @xmath307 . at a given center - of - mass momentum ,
the density of states in @xmath262-space is @xmath308 . to see this ,
consider first a 1d system .
the possible values of the single - particle wave vectors @xmath134 or @xmath309 are separated by @xmath112 .
assume that a pair @xmath310 corresponds to a certain center - of - mass momentum .
the next possible pair of values that corresponds to the same center - of - mass momentum is @xmath311 , which changes @xmath312 by @xmath313 .
conversion into 3d and then into frequency space yields the density of states @xmath314 where @xmath315 .
hence @xmath316 @xmath317 is typically a slowly varying function of @xmath318 .
for large @xmath319 , the phase factor @xmath320 oscillates rapidly and thus the integral is almost zero .
this is the rapid dephasing of the reservoir stats that we mentioned earlier .
the integral over @xmath321 in eq .
( [ rho - tilde - dot-2 ] ) is thus dominated by the values at @xmath322 . this justifies a markov approximation @xcite , which consists of replacing @xmath323 in eq .
( [ rho - tilde - dot-2 ] ) by @xmath324 . since the integral over @xmath321
is dominated by the values at @xmath322 , we can extend the lower bound of the integral to @xmath325 @xcite .
we substitute @xmath326 and eq .
( [ rho - tilde - dot-2 ] ) becomes @xmath327 b_{{\bf kpp}'}(t ) \nonumber \\ & & + h.c.\end{aligned}\ ] ] with @xmath328 we swap the two integrals , performing the time integration first .
we use @xcite @xmath329 where @xmath94 denotes the principal value for the following @xmath330 integration .
hence @xmath331 where @xmath317 in eq .
( [ gamma - d ] ) is to be taken at @xmath332 , which reflects energy conservation .
note that our results for @xmath333 and @xmath334 are twice as large as eqs .
( 1.68 ) and ( 1.70 ) in ref .
this difference is due to the two terms in eq .
( [ delta ] ) . ref .
@xcite considers reservoir states with only one particle , so that only one term occurs in the expression analogous to eq .
( [ delta ] ) .
we transform eq .
( [ rho - tilde - dot-3 ] ) back from the interaction picture to the schrdinger picture .
we use that @xmath333 and @xmath334 are real and obtain @xmath335 = } \\
\nonumber & = & \sum_{{\bf kpp } ' } \frac{\gamma_d}2 \left [ a_{i,{\bf k}+{\bf p } ' } a_{i,{\bf k}-{\bf p ' } } \rho(t ) , a_{i,{\bf k}+{\bf p}}^\dag a_{i,{\bf k}-{\bf p}}^\dag \right ] + h.c.\end{aligned}\ ] ] with @xmath336 @xmath337 is very reminiscent of the energy shift in ordinary second - order perturbation theory , because @xmath337 describes processes in which a pair of particles in state @xmath338 makes a transition to a state @xmath339 with relative momentum @xmath253 and then returns to state @xmath338 .
energy is usually not conserved in such virtual transitions and the energy difference between the initial and intermediate state appears in the denominator of eq .
( [ delta - p ] ) .
this denominator gives the typical weighting for such virtual transitions , known from perturbation theory .
a calculation of the value of @xmath334 from eq .
( [ delta - p ] ) is difficult , even if the incoming particles are ultracold .
this is because , first , eq .
( [ sum - g - fi ] ) can not be used because the energy denominator depends on @xmath340 and @xmath341 and , second , because energy is not conserved in the virtual transitions so that a calculation of the integral requires knowledge of the @xmath11-matrix elements in eq .
( [ g - alpha - beta - ii ] ) over a large range of energies @xmath342 , even if @xmath343 is fixed to be ultracold .
the energy dependence of the @xmath11-matrix elements is sensitive to many details of the true potential .
a calculation of @xmath334 is thus beyond the scope of the present work .
we see that @xmath337 in eq .
( [ h - delta ] ) has exactly the same form as the elastic collision term @xmath344 in eq .
( [ h - sc ] ) . in the ultracold limit
, @xmath337 thus simply contributes to the mean - field energy .
any attempt to measure @xmath345 in the many - body system will always yield the sum of the two contributions , so that we can simply absorb @xmath334 in an effective value of @xmath345 , that needs to be determined experimentally . a calculation of @xmath333 from eq .
( [ gamma - d ] ) is much simpler because it requires knowledge of the @xmath11-matrix elements only for outgoing momenta @xmath253 that satisfy energy conservation .
there is no additional denominator and we can use eqs .
( [ sum - g - fi ] ) , ( [ kappa ] ) and ( [ g ] ) with @xmath346 and @xmath347 .
we obtain @xmath348 for ultracold incoming particles , we use eqs .
( [ a ] ) and ( [ g-3d ] ) and obtain @xmath349 this is independent of @xmath260 , @xmath261 , and @xmath262 , so that eq .
( [ rho - tilde - dot-4 ] ) can be transformed into a simple expression in position representation .
this yields eq .
( [ d ] ) , which is the central result of this appendix .
finally , we discuss how our results relate to the existing literature on scattering theory . to this end , we note that the rate coefficient @xmath350 in eq .
( [ k-3d ] ) can be calculated in an alternative way , based on the observation that the scattering rate per particle in a gas is proportional to @xmath351 , where @xmath352 is the cross section and @xmath13 is the wave vector of the relative motion of the two particles before the collision . for elastic and inelastic two - body collisions of identical bosons
one obtains @xcite @xmath353 respectively .
deriving the rate coefficient @xmath350 from @xmath354 is a delicate issue because of several subtle factors of 2 .
an unambiguous way to handle those factors is based on a quantum boltzmann equation @xcite and confirms our result eq .
( [ k-3d ] ) .
this appendix discusses the applicability of the model presented in sec .
[ sec - loss-1d ] to our experiment reported in ref . @xcite .
first , we note that the derivation of @xmath33 in ref .
@xcite relies on the assumption that the kinetic energy in the relative motion of two particles @xmath355 before a collisions is less than the splitting of the transverse harmonic oscillator levels with even angular momentum @xmath356 .
this is equivalent to @xmath357 . in a fermionized ground state ,
noticeable momenta can occur even at zero temperature .
we can use the @xmath98 to estimate the typical momenta in the ground state .
we note that the @xmath98 are not actually momenta , see _
e.g. _ ref .
@xcite , but the typical width of the momentum distribution is set by @xmath98 .
we are interested in the typical relative momentum in the fermionized ground state , which we can thus estimate from eq .
( [ k - j ] ) as @xmath358 .
the experiment in ref .
@xcite was performed at an initial density of @xmath359 and with @xmath360 nm , so that @xmath361 which is well inside the regime @xmath357 .
second , we mention another assumption in the derivation of @xmath33 in ref .
@xcite , namely that the 1d scattering process can be modeled by a 1d delta potential only for low enough momenta . for @xmath362 , @xmath363 , and @xmath364 @xmath365/s @xcite we
numerically evaluate the analytic results of ref .
we find that the transmission coefficient @xcite for the scattering process is altered by less than 10% when making the transition from the 3d regularized delta potential to the 1d delta potential , thus showing that the model is , indeed , a good approximation .
finally , we briefly discuss the part of our experiment in ref .
@xcite in which a weak lattice is applied along the one dimension .
if two particles occupy the same lattice site , the resulting on - site interaction is described by a matrix element @xmath366 .
we will discuss now , why the expression @xmath367 is valid , where @xmath368 is a wannier function .
first , the lattice potential at a single lattice site is to a good approximation harmonic , with an aspect ratio of less than 3 for the parameters of our experiment .
scattering processes within one lattice site are therefore not in a 1d regime and only for much larger aspect ratios would we obtain @xmath369 , with a 1d wannier function .
second , if two particles occupy the same site , this 3d system at one site has @xmath370 . to see this we consider a lattice depth of @xmath371 @xcite along all three dimensions , where @xmath372 is the recoil energy , yielding a harmonic oscillator length of @xmath373 nm at one site . at @xmath364 @xmath365/s @xcite and @xmath363 , we obtain for two particles at one site @xmath374 .
the fact that @xmath370 means that two particles at an isolated single site are in the weakly correlated regime and the single - particle wannier function can be used to calculate @xmath366 .
we thank paul julienne for enlightening discussions about the subtle factors of 2 in appendix [ app - scattering ] .
we acknowledge financial support of the german excellence initiative via the program nanosystems initiative munich and of the deutsche forschungsgemeinschaft via sfb 631 .
jjgr acknowledges financial support from the ramon y cajal program and the spanish projects fis2006 - 04885 and cam - ucm/910758 . | we show that strong inelastic interactions between bosons in one dimension create a tonks - girardeau gas , much as in the case of elastic interactions .
we derive a markovian master equation that describes the loss caused by the inelastic collisions .
this yields a loss rate equation and a dissipative lieb - liniger model for short times .
we obtain an analytic expression for the pair correlation function in the limit of strong dissipation .
numerical calculations show how a diverging dissipation strength leads to a vanishing of the actual loss rate and renders an additional elastic part of the interaction irrelevant . |
the advent of graphene @xcite opened a door for exploration of new phenomena in two - dimensional dirac - like condensed matter systems .
one of the intriguing questions is of superconducting correlations of electrons on the honeycomb lattice system .
superconductivity has been induced in short graphene samples through proximity effect with superconducting contacts @xcite .
this indicates that cooper pairs can propagate coherently in graphene . in principle superconductivity on the graphene honeycomb lattice
can be induced by short - range attractive interactions and explorations of allowed possibilities were given in refs .
@xcite . among the most interesting
is the so - called @xmath3 superconducting instability introduced in ref .
it would be supported by the most natural nearest - neighbor attractive interaction and have distinct features of the dirac electrons .
later it was showed @xcite , by a restricted ( low - energy ) analysis , that this state may be less energetically favorable with respect to kekule - like order parameter arrangements .
nevertheless , the @xmath4 instability seems , though an exotic state , a very attractive possibility because of its underlying symmetry of the order parameter , the same as for pfaffian quantum hall state @xcite or @xmath3 spinless superconductor @xcite .
the later systems support non - abelian statistics , which is at the heart of the idea of the topological computing @xcite .
there is an important difference between these states and the proposed graphene state
. the superconducting instability in graphene does not break time - reversal symmetry and those systems do .
due to the valley degeneracy we effectively have two @xmath5 order parameters and that requires additional understanding of intertwined correlations and underlying symmetries .
one way , just as in the pfaffian state @xcite , is to look for the ground state wave function and recognize the structures and symmetries . in this paper , in the first part , we will find the effective ( long - distance ) expression for the ground state wave function of the @xmath3 spin - singlet instability described in ref . and display pertinent symmetries in this case .
also a spinless case will be discussed .
we will use the bcs mean - field formalism . in the following section
we will set up the bcs formalism , solve the bogoliubov - de gennes ( bdg ) equations and find the expression for the ground state wave functions .
the last section of the first part is devoted to conclusions .
the second part of the paper is devoted to the @xmath6 superconductivity on the bilayer honeycomb lattice .
we refer reader to this part of the paper for an introduction .
the hamiltonian for free electrons on the honeycomb lattice is @xmath7 where @xmath8 is the hopping energy between nearest neighbor c ( carbon ) atoms , @xmath9 is the on - site annihilation ( creation ) operator for electrons in the sublattice a with spin @xmath10 , and @xmath11 for sublattice b , @xmath12 is the on - site number operator , and @xmath13 is the graphene chemical potential .
we use units such that @xmath14 .
diagonalization of eq.([ho ] ) leads to a spectrum given by : @xmath15 with @xmath16 s defined as @xmath17 , @xmath18 , and @xmath19 , and @xmath20 , @xmath21 is the distance between @xmath22 atoms and @xmath23 is the next to nearest neighbor distance . at the corners of the hexagonal brillouin zone , @xmath24
, we have @xmath25 , and the band has the shape of a dirac cone : @xmath26 where @xmath27
. we will assume the spin - singlet pairing among nearest - neighbors and apply the bcs ansatz with @xmath28 , the superconducting order parameter .
furthermore we assume one and the same @xmath29 for all nearest neighbors , which due to global gauge @xmath30 transformations on @xmath23 s and @xmath31 s can be chosen real and positive @xcite .
the interaction part , @xmath32 , becomes @xmath33 therefore near @xmath34 points @xmath35 , which then describes two @xmath36-wave like superconducting order parameters in a low effective description .
the complete bcs hamiltonian can be now cast in the following form in the momentum space , @xmath37 where @xmath38 with defined @xmath39 and @xmath40 , and , with @xmath41 for short , @xmath42.\ ] ] we look for the solution in the form of a diagonalized bogoliubov bcs hamiltonian , @xmath43 where @xmath44 and @xmath45 , @xmath46 are new quasiparticles at momentum @xmath47 .
for the dispersions we have : @xmath48 where @xmath49 .
we define a general solution @xmath50 as @xmath51 next we have to solve the bogoliubov - de gennes ( bdg ) equations , which follow from the following condition , @xmath52 = e \alpha_{\vec{k}}.\label{beq}\ ] ] from this matrix eigenvalue problem we obtain energies of the bogoliubov quasiparticles , @xmath53 where @xmath54 stands for the particle and hole branches respectively for two kinds of excitations @xmath55 and @xmath56 . for @xmath57 the system is gapless and we need a coupling @xmath58 larger than a critical value for the superconducting instability to exist @xcite .
this can be found considering in the bcs formalism the consistency or gap equation . for each valley
we have to solve the bogoliubov problem using the expansion @xmath59 . near @xmath60
we need to diagonalize the following matrix , @xmath61 , that comes out of eq.([beq ] ) : @xmath62,\ ] ] where @xmath63 .
its eigenvectors ( after normalization ) enter the following expressions for bogoliubov quasiparticles : @xmath64 } } \{[e_{\alpha } - ( \mu - v_f k ) ] ( \sqrt{\frac{k}{k^ { * } } } a_{+ \uparrow } + b_{+ \uparrow } ) + s |k| ( \sqrt{\frac{k}{k^ { * } } } a_{- \downarrow}^{\dagger } + b_{- \downarrow}^{\dagger})\},\ ] ] and @xmath65 } } \{[e_{\beta } - ( \mu + v_f k ) ] ( \sqrt{\frac{k}{k^ { * } } } a_{+ \uparrow } - b_{+ \uparrow } ) - s |k| ( \sqrt{\frac{k}{k^ { * } } } a_{- \downarrow}^{\dagger } - b_{- \downarrow}^{\dagger})\ } , \label{beta1}\ ] ] and quasiholes : @xmath66 } } \{-[e_{\alpha } + ( \mu - v_f k ) ] ( \sqrt{\frac{k}{k^ { * } } } a_{+ \uparrow } + b_{+ \uparrow } ) + s |k| ( \sqrt{\frac{k}{k^ { * } } } a_{- \downarrow}^{\dagger } + b_{- \downarrow}^{\dagger})\},\ ] ] and @xmath67 } } \{-[e_{\beta } + ( \mu + v_f k ) ] ( \sqrt{\frac{k}{k^ { * } } } a_{+ \uparrow } - b_{+ \uparrow } ) - s |k| ( \sqrt{\frac{k}{k^ { * } } } a_{- \downarrow}^{\dagger } - b_{- \downarrow}^{\dagger})\},\label{beta2}\ ] ] for the bogoliubov solution near point @xmath68 , where we denoted @xmath69 and @xmath70 .
the natural eigenstates of chirality appeared in our expressions .
for example @xmath71 represents spinor : @xmath72,\label{spinor}\ ] ] which is the eigenstate of the chirality operator @xmath73 , defined with @xmath74 pauli matrices , i.e. the pseudospin ( due to two sublattices ) is along the momentum vector .
the state @xmath75 represents the same spinor because of the interchanged roles of sublattices in the @xmath76 point . to see this in more details
we would like to remind the reader that instead of the dirac free electron representation by the spinor @xmath77 and the chirality operator is defined as @xmath78,\label{chiop}\ ] ] in the bdg formalism we work with @xmath79 note the reversed order of sublattices and the change of the sign of the momentum @xmath47 near @xmath76 point in the bdg formalism with respect to the free one .
thus the lower @xmath80 matrix on the diagonal of the hamiltonian matrix in the free dirac case can be read off from : @xmath81 \left[\begin{array}{cc } - \mu & - v_f k^ * \\ - v_f k & - \mu \end{array } \right ] \left[\begin{array}{c } b_{\vec{k}_{- } + \vec{k } , \sigma}\\ a_{\vec{k}_{- } + \vec{k } , \sigma } \end{array } \right],\label{drep}\ ] ] i.e. it is equal to @xmath82 .
note that if we change the sign of @xmath47 vector in eq.([drep ] ) i.e. @xmath83 the off - diagonal elements in the matrix will change the sign , so that in this basis in the free representation the chirality operator will not have minus sign in the lower right entry of the matrix representation in eq.([chiop ] )
. therefore @xmath84 represents the same spinor ( up to a phase factor ) as in eq.([spinor ] ) and the same chirality eigenstate ( with positive eigenvalue ) as we pointed out earlier .
nevertheless in the bogoliubov representation we still have @xmath85 \left[\begin{array}{cc } \mu & - v_f k^ * \\ - v_f k & \mu \end{array } \right ] \left[\begin{array}{c } a_{- \downarrow}^{+}\\ b_{- \downarrow}^{+ } \end{array } \right],\ ] ] i.e. the matrix is @xmath86 , and the representation of the chirality operator stays the same as in eq.([chiop ] ) .
we will use this fact later . on the other hand the combinations in eqs .
( [ beta1 ] ) and ( [ beta2 ] ) : @xmath87 and @xmath88 have the pseudospin vector in the opposite direction of the momentum vector @xmath47 .
it is thus natural to introduce the following notation : @xmath89 where @xmath90 and @xmath91 denote the chirality i.e. whether the pseudospin vector is along or in the opposite direction with respect to the @xmath47 vector , respectively .
we have to note that these electron operators are defined up to a phase factor , most importantly @xmath92 phase .
this degree of freedom should not influence the physics , but we chose the definitions so that later the symmetry under exchange of particles in the ground state wave function is transparent .
the @xmath50 and @xmath93 sectors are obviously decoupled in the bogoliubov description and we can concentrate and closely examine the @xmath50 sector first .
furthermore we do not have to consider @xmath76 point separately as the symmetry considerations tell us that the bdg equations around this point will induce the coupling or states of an electron around @xmath94 point with @xmath95 projection of spin and those around @xmath76 point with @xmath96 projection of spin .
thus it suffices to consider @xmath50 sector first ( with @xmath97 and @xmath98 ) and then use the symmetry arguments , more precisely antisymmetry under real spin exchange to recover the whole ground state wave function .
we can rewrite @xmath50 s in the following form , @xmath99 we should demand @xmath100 and @xmath101 , for any @xmath102 , if @xmath103 is to represent the ground state vector .
that implies that in the @xmath50 sector of @xmath94 point we have the following contribution to the ground state , @xmath104 where @xmath105 denotes the vacuum .
this state is annihilated with both , @xmath106 and @xmath107 the symmetry arguments demand that we should get a similar expression considering bdg equations at @xmath76 point .
if we denote by @xmath108 , the ground state vector in the @xmath50 sector should look like : @xmath109 + \frac{g_{\alpha}^{2}(k)}{2 } [ c_{+ \uparrow v}^{\dagger } c_{- \downarrow v}^{\dagger } + c_{- \uparrow v}^{\dagger } c_{+ \downarrow v}^{\dagger}]^{2}\}|0\rangle \nonumber \\ = & & \exp\ { \sum_{k } g_{\alpha}(k ) [ c_{+ \uparrow v}^{\dagger } c_{- \downarrow v}^{\dagger } + c_{- \uparrow v}^{\dagger } c_{+ \downarrow v}^{\dagger}]\}|0\rangle\end{aligned}\ ] ] now we can identify @xmath110 to represent a fourier transform of the wave function of a cooper pair of electrons , which is a spin - singlet with respect to spin degree of freedom and a triplet state ( symmetric under exchange ) with respect to valley @xmath111 degree of freedom . if we defined differently our electron operators there would be possibility for @xmath110 to acquire the phase factor @xmath92 , which would make the identification of the antisymmetry under exchange harder . taking into account the @xmath93 sector ( with the chirality in the opposite direction of the momentum : @xmath91 ) the complete ground state vector is @xmath112 + \sum_{k } g_{\beta}(k ) [ c_{+ \uparrow w}^{\dagger } c_{- \downarrow w}^{\dagger } + c_{- \uparrow w}^{\dagger } c_{+ \downarrow w}^{\dagger}]\}|0\rangle , \label{gsv}\ ] ] where @xmath113 using the long - distance ( low - momentum ) expansions for @xmath114 and @xmath115 , for finite @xmath13 , @xmath116 we find the long - distance behavior of the pair wave function to be @xmath117
i.e. we have a case for a weak pairing @xcite . as emphasized in ref .
the term weak pairing does not mean also weak coupling , it stands for a phase with an unusual large spread of the cooper pairs . on the other hand for @xmath57
we have that @xmath110 and @xmath118 are two constants and the cooper pairs are localized on a short scale @xmath119 in the graphene system at the neutrality point .
thus for @xmath57 we have a case for a strong pairing .
the ground state vector ( wave function ) in eq.([gsv ] ) displays two kinds of cooper pairs , each antisymmetric under combined exchange of ( a ) orbital , ( b ) valley ( @xmath120 ) , and ( c ) spin @xmath121 degree of freedom .
two kinds of cooper pairs stem from the chirality ( sublattice ) degree of freedom intimately connected with the dirac - nature of the electron with both , particles and holes .
they both , particles ( with positive chirality @xmath90 at @xmath94 ) and holes ( with negative chirality @xmath91 at @xmath94 ) , constitute cooper pairs , which are symmetric under @xmath122 transformation . in the long distance limit
we recover the form of the wave function of ordinary @xmath123-wave superconductor as given in ref .
, though with more , two - component , degrees of freedom . the cooper pair wave function is antisymmetric under spin exchange and symmetric under exchange of valley ( @xmath120 ) , sublattice @xmath124 , and orbital degrees of freedom .
next we will discuss the spin - triplet case , more precisely we will assume that the system is spin - polarized and not consider spin in the following
. therefore fermions are spinless just like in the pfaffian case , but they live on the honeycomb lattice .
we will assume @xmath125 .
in this case the bogoliubov problem in eq.([hbcs ] ) for the spin - singlet pairing transforms into a similar one with @xmath126 and @xmath127 , and the matrix @xmath128 becomes as follows @xmath129.\ ] ] around the @xmath94 point we have @xmath130,\ ] ] where @xmath131 as before .
the problem around the @xmath76 point is a copy of the problem around the @xmath94 point .
now the @xmath128 matrix around @xmath94 point can not be cast , as in the spin - singlet case , in the following form , @xmath132,\ ] ] where @xmath133 is the @xmath134 identity matrix , which commutes with the chirality matrix ( eq.[chiop ] ) .
@xmath128 around @xmath94 point can be compactly written as @xmath135,\ ] ] and it does not commute with the chirality operator .
the eigenstates of the bogoliubov problem do not have to be the eigenstates of chirality .
we find the following eigenvalues @xmath136 , where @xmath137 and @xmath138 are two branches as before .
the associated eigenvectors can be written as sums of fermionic particle eigenstates of chirality only in the low - momentum limit and we list those connected with positive eigenvalues , @xmath139,\ ] ] and @xmath140,\ ] ] and negative eigenvalues , @xmath141,\ ] ] and @xmath142.\ ] ] similarly as before we can define @xmath143 and the ground state vector can be cast in the following form , @xmath144 in this case each cooper pair is antisymmetric under exchange of @xmath120 points i.e. valley degree of freedom and symmetric under exchange of sublattices i.e. chirality @xmath145 .
depending on our definitions for @xmath146 s two degrees of freedom can exchange the symmetry properties .
we find again the weak pairing @xmath147 behavior in the orbital part .
we derived the ground state wave functions for the superconductivity on the honeycomb lattice induced by nearest - neighbor attractive interactions and with order parameter independent of the direction to any of the nearest neighbors . although the order parameter in momentum space has the @xmath148 form in a low effective description the cooper pair wave function behaves as @xmath123-wave ( with no angular dependence ) and decays as @xmath149 .
other ( discrete ) degrees of freedom combine to make the cooper pair antisymmetric under exchange . at the point of the transition , @xmath150 , in the spin - singlet case
a strong pairing ( of the order of lattice spacing ) occurs .
topological superconductors in a strict sense or what we also call non - trivial topological superconductors have odd number of majorana modes moving in each direction on the edge of such a superconductor @xcite . in the case of trivial topological superconductors we have
even number of majorana modes i.e. by combining them in pairs we can talk about dirac fermions on their edge .
@xmath123-wave superconductor in two dimensions is always topological in the sense that it has a gap in its bulk and non - trivial degeneracy of the ground state on the torus ( equal to four ) @xcite .
we have to use one bose field ( one dirac fermion ) to describe the edge of such a system @xcite . on the other hand
if we combine two @xmath36-wave superconductors , with @xmath1 and @xmath2 orbital symmetry , and each of the two corresponds to one let s say definite projection of spin we have the case for a non - trivial topological superconductor
. in that case on the edge live two majorana modes that are moving in opposite directions and each is associated with different projection of electron spin .
this represents a
helical " edge where we have a pair of edge majorana modes ( moving in opposite directions ) that are connected with a time - reversal operation .
this is the simplest topological
superconductor we can imagine in two dimensions and has yet to be realized and detected in experiments . in three dimensions
a realization of topological superconductor is he@xmath0 b - phase @xcite . on the other hand the honeycomb lattice ,
nowadays very much connected with the research on graphene , is a playing ground for various , among others topological , phases .
the first topological insulator was introduced on the honeycomb lattice with a special interaction @xcite . while considering possibilities for superconducting instabilities on the honeycomb lattice and graphene in ref .
, a phase was proposed with two @xmath36-wave order parameters ( each near two effective descriptions in @xmath47-space i.e. two valleys ) . though one might expect that , while considering triplet pairing i.e. if we suppress spin , this would lead to a non - trivial topological superconductor with a pair of majorana modes ,
this is not the case as we demonstrated in the first part of the paper ( ref . ) .
we found that the ground state wave function is antisymmetric with respect to the valley degree of freedom and that the orbital part of the cooper pair is with no angular dependence i.e. a @xmath123-wave . as we emphasized earlier in the case of @xmath123-wave
we expect one dirac fermion per degree of freedom on the edge i.e. no non - trivial behavior .
in this paper we will derive ground state wave functions for some superconducting instabilities that may emerge due attractive interactions on a two layer system in which each layer represents a honeycomb lattice .
the two lattices are stacked as in the bilayer graphene i.e. in the way of bernal stacking . as in the bilayer graphene
, we expect that the pseudospin vector connected with sublattice degrees of freedom will not follow the momentum vector in a parallel or antiparallel fashion , like on ordinary honeycomb lattice or graphene , but rotate for a whole angle as a result of a rotation of the momentum vector for a half an angle .
this feature of free electrons on the bilayer honeycomb lattice will reflect in the description of cooper pairs when attractive interactions are introduced .
the explicit ground state wave functions and cooper pair structure will help us to see more closely the nature of pairing in this system . we will find the @xmath36-wave angular dependence in the orbital part . in the following section , we will formulate bogoliubov - de gennes ( bdg ) equatons for this system . in the next section
the explicit solutions with corresponding ground state wave functions will be given in the case of ( a ) spin - singlet and ( b ) spinless ( spin - triplet ) pairing .
then we will examine whether these systems are truly gapped in the bulk , and , in the spin - singlet case , its edge spectrum .
the last section is devoted to discussion and conclusions .
the hamiltonian for free electrons on two honeycomb lattices , which are bernal stacked , is @xmath151 the index @xmath152 denotes the layer index . in fig .
1 the relative positions of two triangular sublattices , @xmath153 and @xmath154 , for the lattice 1 , and @xmath155 and @xmath156 , for the lattice 2 are illustrated . in eq.([freeham ] ) @xmath8 is the hopping energy between nearest neighbor c ( carbon ) atoms in the case of the bilayer graphene in each layer , and @xmath157 is the same energy for hopping between the layers .
the on - site creation ( annihilation ) operators , @xmath158 , are for the electrons in the sublattice @xmath159 of the layer @xmath160 with spin @xmath10 , and @xmath161 for the electrons in the sublattice @xmath162 , @xmath163 is the on - site number operator , and @xmath13 is the chemical potential .
@xmath16 s are defined as @xmath164 , @xmath18 , and @xmath165 , and @xmath166 , @xmath21 is the distance between @xmath22 atoms and @xmath23 is the next to nearest neighbor distance .
we use units such that @xmath14 . by introducing fourier transforms @xmath167 and @xmath168 etc . and diagonalizing the hamiltonian we find for the spectrum , @xmath169 where @xmath170 , and @xmath171 stand for two kinds of branches .
near @xmath172 points , the corners of the hexagonal brillouin zone , @xmath173 , we have @xmath174 and in the limit @xmath175 the lower positive and higher negative branch have the folowing dispersion relation , @xmath176 where @xmath177 and @xmath178 , the fermi - dirac velocity .
the effective hamiltonian near @xmath172 points @xcite is @xmath179,\ ] ] and acts on the subspace of ( pseudo)spinors @xmath180^{t}\ ] ] around point @xmath94 , and @xmath181^{t}\ ] ] around point @xmath76 .
@xmath182 can be rewritten as @xmath183 = -\frac{\vec{k}^2}{2 m^ * } \vec{\sigma } \vec{n},\ ] ] where @xmath184 and @xmath185 , and @xmath186 s are pauli matrices .
the operator @xmath187 encodes the projection of the pseudospin on direction @xmath188 . for eigenstates as @xmath189,\label{sps } \;\ ; { \rm and}\;\ ; \chi^- = \left[\begin{array}{c } \frac{k^{*}}{k } \\ -1 \end{array } \right],\ ] ] the direction @xmath188 may be interpreted as the direction of the pseudospin vector , with projection ( chirality ) equal to @xmath190 in the case of @xmath191 , and @xmath192 in the case of @xmath193 .
thus in the case of these eigenstates we see explicitly our previous remark that the pseudospin vector rotates for an angle while @xmath47 vector rotates for half an angle circling the fermi surface around @xmath172 points .
that feature of the solutions of the free problem leads to non - trivial pairing in the orbital part of cooper pairs as we will see later .
this is to be contrasted to the behavior in the monolayer , a single honeycomb lattice , where the rotation of @xmath47 vector is strictly followed by the rotation of the pseudospin vector .
it is accompanied by @xmath123-wave pairing , when special ( nearest - neighbor ) attractive interactions are applied . in that case
although two order parameters are of , @xmath1 , and @xmath2 type we have the trivial ( @xmath123-wave ) behavior in the orbital part as we have shown earlier .
as the reader may have noticed we did not include the direct hopping between the atoms of b1 and b2 sublattice .
this inclusion is required when we model bilayer graphene @xcite , but even there for realistic parameters this does not influence the physics at high electron momenta or strong magnetic fields @xcite .
but we will consider nearest - neighbor attractive interactions between electrons on b1 and b2 sublattice .
namely these sublattices by themselves make a honeycomb lattice as we can verify by looking at fig
. 1 . due to the strong hopping between a1 and a2 sublattice the complete low - energy physics
is projected onto b1 and b2 sublattice .
if the interactions are not too strong they can be simply added to this low - energy subspace .
the on - site repulsive interactions can be introduced and we do not expect that will change our conclusions . therefore the complete hamiltonian will include nearest - neighbor attractive interactions between electrons on b1 and b2 sublattice as follows , @xmath194 where @xmath195 .
we will assume the spin - singlet pairing among nearest neighbors and apply the bcs ansatz with @xmath196 the superconducting order parameter .
furthermore we assume one and the same @xmath197 for all nearest neighbors , which due to global gauge ( u(1 ) ) transformations on @xmath198 s and @xmath199 s can be chosen real and positive .
the interaction part , @xmath200 , becomes @xmath201 therefore near @xmath172 points @xmath202 , which then describes two @xmath36-wave like superconducting order parameters in a low - energy effective description .
taking into account the complete low - energy reduction the total bcs hamiltonian can now be cast in the following form in the momentum space near @xmath94 , @xmath203 , @xmath204 where @xmath205 and , with @xmath206 , @xmath207.\ ] ] we will omit the discussion concerning @xmath208 in the neighborhood of @xmath76 and momenta : @xmath209 .
this entails operators which combine @xmath96 spin with momenta @xmath210 and @xmath95 spin with momenta @xmath211 , and will not provide any new information for the structure of the ground state wave function or energy dispersion at small momenta .
we can simply include these operators at the end in the ground state wave function following symmetry requirements for the spin - singlet pairing .
we look for the solution of eq .
( [ hbcs2 ] ) in the form of a diagonalized bogoliubov bcs hamiltonian , @xmath43 where @xmath44 and @xmath45 , @xmath46 are new quasiparticles at momentum @xmath47 . for the dispersions
we have : @xmath48 where @xmath49 .
we define a general solution @xmath50 as @xmath212 next we have to solve the bogoliubov - de gennes ( bdg ) equations , which follow from the following condition , @xmath52 = e \alpha_{\vec{k}}.\label{beq2}\ ] ] we need to diagonalize the following matrix , @xmath61 , that comes out of eq .
( [ beq2 ] ) : @xmath213.\ ] ] from this matrix eigenvalue problem we obtain energies of the bogoliubov quasiparticles , @xmath214 where @xmath54 stands for the particle and hole branches respectively for two kinds of excitations @xmath55 and @xmath215 .
the eigenvectors of matrix @xmath61 ( after normalization ) enter the following expressions in the long - distance limit for bogoliubov quasiparticles : @xmath216 and @xmath217 and quasiholes : @xmath218 and @xmath219 for the bogoliubov solution near point @xmath68 , where we denoted @xmath220 and @xmath221 .
it is helpful to introduce the creation and annihilation operators of states of definite chirality or projection of the pseudospin along the @xmath188 vector ; we denote by @xmath90 positive projection , and by @xmath91 negative projection : @xmath222 notice that the states are defined up to a phase factor according to the definitions in eq.([sps ] ) , and that in the states around @xmath76 point the roles of electrons in different layers are interchanged
. we may then define their superpositions , @xmath223 then we can rewrite bogoliubov operators as : @xmath224 with @xmath225 , and find that the wave function , @xmath226 , @xmath227 is annihilated by quasiparticle annihilation operators , @xmath228 , and by quasihole creation operators , @xmath229 .
therefore the new ground state is @xmath230 where we explicitly introduced the sector that couples momenta around @xmath94 with spin @xmath95 and momenta around @xmath76 with spin @xmath96 by enforcing the explicit spin - singlet pairing which we introduced at the beginning . from the structure of the ground state wave function for spin - singlet pairing in eq.([gs1 ] )
we find that electrons pair between @xmath94 and @xmath76 point have the same pseudospin and therefore we have two distinct cooper pairings for two orthogonal pseudospin states , which we denoted by @xmath231 and @xmath232 .
each pair has a @xmath36-wave pairing in the orbital part and , as we work with a time - reversal invariant system , two distinct cooper pairings are accompanied by two distinct , @xmath1 and @xmath2 , symmetries in the orbital part .
each cooper pair is antisymmetric under spin exchange , valley exchange , and exchange in the orbital part and symmetric under sublattice ( pseudospin ) exchange .
therefore the ground state wave function in eq.([gs1 ] ) describes a cooper - paired collection of fermions -electrons on the bilayer honeycomb lattice with unconventional @xmath36-wave pairing .
the non - trivial ( non-@xmath123-wave ) pairing can not be eliminated with a gauge transformation . to preserve the form of the free part of the hamiltonian
any gauge transformation should be @xmath233 and @xmath234 , and @xmath235 and @xmath236 , and if we rewrite @xmath237 in terms of these operators @xmath238 we see that by this gauge transformation we can not eliminate simultaneously the angular dependence in the two types of cooper @xmath36-wave pairings .
next we will discuss the spinless case .
we will assume that all electron spins are polarized and that @xmath239 .
in this case the bogoliubov problem in eq.([hbcs2 ] ) for the spin - singlet pairing transforms into a similar one with @xmath240 and @xmath241 and the matrix @xmath208 becomes as follows , @xmath242.\ ] ] the problem around the @xmath76 point is a copy of the problem around @xmath94 .
we find the following eigenvalues for the eq.([beq2 ] ) , @xmath243 where @xmath54 stands for the particle and hole branches respectively for two kinds of excitations @xmath55 and @xmath215 .
the eigenvectors enter the following expressions in the long - distance limit for bogoliubov quasiparticles : @xmath244 and @xmath245 and quasiholes : @xmath246 and @xmath247 introducing as in the spin - singlet case the following pseudospin operators : @xmath248 we can rewrite the eigenvectors as follows , @xmath249 similarly as in the spin - singlet case we can find that the ground state wave function can be expressed as @xmath250 the cooper pairs are symmetric under valley and sublattice ( pseudospin ) exchange and antisymmetric under exchange in the orbital part .
we have two kinds of cooper pairs with underlying @xmath251 and @xmath2 symmetry .
in ref . the case of spin - singlet pairing in the monolayer was thoroughly discussed .
a topological phase structure was described with four ( due to valley and spin ) dirac edge modes .
this is consistent with our previous calculations of the ground state wave function that has explicit @xmath123-wave dependence @xmath252 for @xmath253 ( chemical potential ) .
here we extended ground state calculations to the spin - singlet and spin - triplet case of the bilayer .
it is appropriate to ask the question raised in ref . for spin - singlet and spin - triplet monolayer case : are these phases truly gapped or there are nodes for some @xmath102 s in the bulk spectrum ? in the same reference
it was found that in the spin - triplet case , as opposed to the spin - singlet case , there are nodes in @xmath102 space at which gap is equal to zero .
we will find a similar situation in the bilayer case , except that in the case of spin - singlet pairing there is a critical value for chemical potential above which we have a truly gapped - topological phase .
( the triplet case just as in the monolayer analysis in ref . has nodes in the bulk spectrum . )
it is not hard , by repeating the approach of ref .
, to find expressions for the matrix , @xmath254 , that enters bdg equations in both cases for general ( not low ) momentum @xmath102 .
they are @xmath255\label{mat},\ ] ] where @xmath256 and @xmath257 and @xmath258 stand for spin - singlet and spin - triplet case respectively .
we look for the zeros of bogoliubov quasiparticle energies , expressed in the low @xmath102 limit in eq.([e1 ] ) and eq.([e2 ] ) , when @xmath259 . with no low momentum limit , from eq.([mat ] ) , their expressions are , @xmath260}},\ ] ] where we used shorthand notation taking @xmath261 , and @xmath262 and @xmath263 correspond to the spin - singlet and spin - triplet case respectively .
we assume that @xmath264 is small with respect to @xmath13 so that possible nodes can be only near fermi surface defined by @xmath265 .
equating @xmath262 and @xmath263 to zero is equivalent to the following condition , @xmath266 where @xmath267 is the phase of @xmath268 - complex number in general . if we assume that we work with @xmath102 s near fermi surface we have approximately @xmath269 where @xmath270 is defined by @xmath271 as a small depature from the fermi surface value .
therefore we can approximate that for possible nodes near fermi surface in the case of spin - singlet pairing @xmath268 is imaginary i.e. @xmath272 and in the case of spin - triplet pairing @xmath268 is real i.e. @xmath273 .
it is not hard to find nodes in the spin - triplet case . from the definition ,
@xmath274 we can recognize that possible positions of nodes can be restricted to @xmath275 , because in that case @xmath268 takes real values ; @xmath276 . then in our approximation from eq.([cond ] ) and definition @xmath277 we have @xmath278 and that with @xmath275 defines a position of a single node .
therefore the existence of this node ( and other related by symmetry ) tell us that this phase is likely to be gapless even in the bulk and can not represent a topological phase . in the spin - singlet case , if we rescale the momentum @xmath279 as @xmath280 in eq.([def ] ) , the condition that @xmath268 is purely imaginary demands that @xmath281 then @xmath282 expressed differently as @xmath283 this leads to the conclusion that for large enough @xmath284 i.e. chemical potential this equation does not have a solution for @xmath285 .
we find that for @xmath286 no solution exists .
therefore for large enough chemical potential we can have a spin - singlet topological phase i.e. a phase with no gapless bulk excitations . to further examine the topological nature of the spin - singlet phase we will derive its edge modes . with respect to the lattice structure we will consider a particular geometry where the system , defined on a half - plane , has the edge at @xmath287 .
we remind the reader that we use the convention in which @xmath120 vectors are along @xmath288 axis .
this choice of boundary corresponds to so - called armchair boundary condition for which we require that the solutions of bdg equations vanish at @xmath287 .
we will consider bdg equations in the low @xmath102 limit around @xmath120 points , neglect terms quadratic in @xmath289 , and employ the substitution @xmath290 to get their form in the real space ; due to the symmetry of the problem we seek solutions in the form @xmath291 and keep the @xmath279 dependence .
the expression for bdg matrix at momentum @xmath292 is @xmath293,\ ] ] and at momentum @xmath294 , @xmath295,\ ] ] and we immediately see a reduction of the problem to four sets of equations with the substitution @xmath290 . if we denote by @xmath296 a form of a solution , where we used @xmath96 and @xmath95 to denote in a shorthand notation _
( compare with eq.([gensol ] ) where the same symbols were used for real spin ) , we have four sets of equations of similar form . for example , one set at @xmath60 point is : @xmath297 just as argued in ref . in the case of a simple @xmath36-wave superconductor ,
when @xmath298 and @xmath299 we have a zero mode , @xmath300 .
the difference here is that @xmath301 and @xmath302 carry opposite valley indexes and we can not construct majorana bogoliubov quasiparticle with them , but only a dirac one . for @xmath303 we find a chiral ( unidirectional ) mode : @xmath304 . considering analogous equation for @xmath305 and @xmath302 at @xmath76 point we get an additional solution with the same chirality ( the same direction of @xmath279 ) , but these two solutions enter the boundary condition which requires that @xmath306 vanishes at the boundary .
therefore one solution with @xmath307 is @xmath308 where @xmath309 . the other , with @xmath310 and @xmath311
, we find in analogous way .
therefore , we find , as expected from the ground state wave function in eq.([gs1 ] ) , two ( dirac ) solutions with opposite directions of motion along the edge that carry opposite pseudospin .
in addition , due to the spin degree of freedom we expect doubling of modes i.e. in total two dirac modes in each direction .
we do not have to go into a discussion of topological invariants to recognize that ground states for spin - singlet pairing , eq .
( [ gs1 ] ) , and for spin - triplet pairing , eq.([gs2 ] ) , may represent ground states of trivial topological superconductors ; spin and valley degree of freedom induce the doubling of majorana modes that may follow @xcite from @xmath36-wave pairing in the orbital part .
nevertheless these models of superconductors are interesting in their own right due to the presence of the @xmath36-wave pairings in the orbital part . in the case of spin - singlet pairing
we found that there are no gapless bulk excitations for large enough chemical potential and this phase can represent a trivial topological superconductor . in deriving ground state wave functions , in the spin - singlet and spin - triplet case
, we assumed validity of small @xmath102 expansion around @xmath312 . in a strict sense
this requires that @xmath313 is in the same neighborhood where we can approximate dispersion relations i.e. either effective hopping ( fermi velocity ) is large or chemical potential is small .
we also allowed the possibility that minima or even nodes ( in the spin - triplet case ) in the spectra of the superconductors can be away from @xmath312 , around @xmath313 . in the cases of the trivial topological superconductors , on the monolayer and bilayer honeycomb lattice , this seems completely justified in the view of their edge spectrum ( see also @xcite ) . in the cases of gapless spin - triplet superconductors derived ground state wave functions
, though we are inclined to associate them with topological phases , may describe these critical states @xcite .
we find that only if we apply bias to the bilayer larger than its chemical potential we can not apply the bdg program around @xmath312 in the way we described in the paper .
( we remind the reader that with the bias the energy minima of the noninteracting problem shift from @xmath312 at @xmath172 points to nonzero @xmath102 s . ) the attractive interactions that we need for the realization of the paired ground states and phases may well be within the reach of future experiments . in the case of a single honeycomb lattice the interactions may be induced by chemically doping the graphene via metal coating @xcite or trapping fermionic atoms in a honeycomb optical lattice @xcite . to conclude , in the second part of this paper we derived ground state wave functions for the superconductivity on the bilayer honeycomb lattice ( with strong interlayer coupling ) induced by attractive interactions between sites that participate in a low - energy description .
as is well - known , without these interactions , free electrons are described by a dirac equation with a quadratic dispersion .
this unusual feature , similarly to @xmath314he - b phase , leads to the description with two kinds of cooper pairs , with @xmath1 and @xmath2 pairing , in the presence of the attractive interactions .
this is expressed in eq.([gs1 ] ) in the case of the spin - singlet pairing . due to the spin degree of freedom
we find doubling of two chiral dirac modes with opposite pseudospin on the edge of this spin - singlet superconductor - a trivial topological superconductor .
this work was supported in part by the ministry of science and technological development of the republic of serbia , under project no .
on171017 . 9 k.s .
novoselov , a.k .
geim , s.v .
morozov , d. jiang , y. zhang , s.v .
dubonos , i.v .
georgieva , and a.a .
firsov , science * 306 * , 666 ( 2004 ) .
heersche , p. jarillo - herrero , j.b .
oostinga , l.m.k .
vandersypen , and a.f .
morpurgo , nature * 446 * , 56 ( 2007 ) .
b. uchoa and a.h .
castro neto , phys .
lett . * 98 * , 146801 ( 2007 ) .
a.m. black - schaffer and s. doniach , phys .
b * 75 * 134512 ( 2007 ) .
b. roy and i. herbut , phys .
b * 82 * , 035429 ( 2010 ) .
g. moore and n. read , nucl.phys .
b * 360 * , 362 ( 1991 )
. n. read and d. green , phys .
b * 61 * , 10267 ( 2000 ) . c. nayak , s.h . simon , a. stern , m. freedman , and s. das sarma , rev .
phys . * 80 * , 1083 ( 2008 ) . for an analysis of possible order parameters and solutions
see d. poletti , c. miniatura , and b. gremaud , arxiv:1006.3179 j.r .
schrieffer , _ theory of superconductivity _ ,
p. 42 , ( addison - wesley , 1988 ) .
hughes , s. raghu , and s .- c .
zhang , phys .
102 * , 187001 ( 2009 ) .
hansson , v. oganesyan , and s.l .
sondhi , ann . phys . *
313 * , 497 ( 2004 ) .
c.l . kane and e.j .
mele , phys .
lett . * 95 * , 226801 ( 2005 ) m.v .
milovanovi , journal of research in physics , novi sad , _ in press _
e. mccann and v.i .
falko , phys .
* 96*,086805 ( 2006 ) .
novoselov , e. mccann , s.v .
morozov , v.i .
falko , m.i .
katsnelson , u. zeitler , d. jiang , f. schedin , a.k .
geim , nature physics * 2 * , 177 ( 2006 ) .
d.l . bergman and k.
le hur , phys . rev.b
* 79 * , 184520 ( 2009 ) .
this may be compared with the case of pfaffian , which can be associated with the transition between 331 state and fermi liquid state in the context of fractional quantum hall effect as in z. papi , m.o .
goerbig , n. regnault , and m.v .
milovanovi , phys .
b 82 , 075302 ( 2010 ) .
zhu , b. wang , and l .- m .
duan , phys .
* 98 * , 260402 ( 2007 ) . | we derive ground state wave functions of superconducting instabilities on the honeycomb lattice induced by nearest - neighbor attractive interactions .
they reflect the dirac nature of electrons in the low - energy limit . for the order parameter that is the same irrespective of the direction to any of the nearest neighbors we find weak pairing ( slowly decaying ) behavior in the orbital part of the cooper pair with no angular dependence . at the neutrality point , in the spin - singlet case
, we recover a strong pairing behavior .
we also derive ground state wave functions for the superconductivity on the bilayer honeycomb lattice , with strong interlayer coupling , induced by attractive interactions between sites that participate in a low - energy description . without these interactions ,
free electrons are described by a dirac equation with a quadratic dispersion .
this unusual feature , similarly to @xmath0he - b phase , leads to the description with two kinds of cooper pairs , with @xmath1 and @xmath2 pairing , in the presence of the attractive interactions .
we discuss the edge modes of such a spin - singlet superconductor and find that it represents a trivial topological superconductor . |
this presentation is based on the paper [ ] .
our goal is to use the saturation / color glass condensate ( cgc ) formalism ( see [ ] for reviews ) to calculate the single transverse spin asymmetry ( stsa ) in polarized proton proton and polarized proton nucleus collisions . the observable is defined by @xmath4 where the arrow indicates spin - up and spin - down outgoing hadron with the spin direction taken here to be along the @xmath5-axis ( with @xmath6 the collision axis ) . for simplicity we will consider scattering of a transversely polarized quark on an unpolarized proton or nuclear target @xmath7 .
the realistic case of the incoming transversely polarized proton can be recovered if one convolutes the cross section we obtain with the polarized proton wave function squared , or , perhaps equivalently , with the proton transversity distribution .
figure [ recoil ] demonstrates that to capture the spin - dependent effects , such as the stsa , one has to go beyond the eikonal accuracy in the calculation .
recoilless eikonal scattering , as shown in the left panel of , is spin - independent and can not generate the asymmetry .
naive inclusion of sub - eikonal corrections is possible by introducing recoil in the scattering , as shown in the right panel of : such diagrams are suppressed by a power of the center of mass energy squared @xmath8 of the collision and are usually discarded in the saturation / cgc formalism since they are very small .
a much larger contribution to stsa may arise from the non - eikonal splitting of the projectile quark into a quark and a gluon , @xmath9 , which is suppressed only by a power of the strong coupling @xmath10 . in the lcpt language
the @xmath1 splitting may take place either long before or long after the interaction with the target , as shown in below .
( splitting during the interaction with the target is suppressed by powers of energy [ ] . )
both such contributions would be included in the light - cone wave function squared of the projectile ( quark ) .
the spin - dependence by itself is insufficient to generate the stsa : one also needs a relative phase between the amplitude and the complex conjugate amplitude contributing to the stsa - generating part of the process [ ] . in the standard interpretation of the sivers effect [ ] both the spin - dependence and the relative phase originate in the wave function squared ( though , often - times , the actual diagrammatic representation of the effect requires a final - state interaction with the polarized projectile [ ] , which , in the cgc rapidity - ordered language , can still be incorporated as a final - state light - cone wave function [ ] ) .
the lowest - order @xmath9 splitting from is too basic to generate the phase . in the collins mechanism
the phase is generated in the fragmentation function [ ] : this can certainly take place in the saturation / cgc framework as well , but is not the main aim of this investigation .
below we show that it is also possible to generate the relative phase in the integration with the unpolarized target .
we thus obtain a new mechanism for generating stsa , where the spin dependence comes from the polarized projectile wave function , while the phase arises in the interaction with the target .
the diagrams for the scattering process of a transversely polarized quark on a proton or nuclear target are shown in .
squaring the sum of these diagrams gives us the graphs shown in which contribute to the inclusive quark production cross section .
the calculation proceeds along the standard steps employing the light - cone perturbation theory ( lcpt ) ( see [ ] for a review ) : the @xmath1 splitting contributes to the light - cone wave function , which factorizes from the interaction with the target .
the general result for the quark production in the @xmath11 scattering given by diagrams in reads @xmath12 with @xmath13 denoting the light - cone wave function squared and @xmath14 representing the interaction with the target with all the coordinate labels shown explicitly in . here
[ eq - kinematic constraints ] @xmath15 are the transverse coordinates of the incoming quark in the amplitude and in the complex conjugate amplitude , while @xmath16 is the fraction of the incoming quark s plus momentum component @xmath17 carried by the produced quark with the plus momentum component @xmath18 , such that @xmath19 .
the light - cone wave function squared @xmath13 can be decomposed into a polarization - independent term and another term which is proportional to the quark transverse polarization eigenvalue @xmath20 , @xmath21 where a straightforward calculation yields [ ] @xmath22\end{aligned}\ ] ] for the unpolarized component and @xmath23 \end{aligned}\ ] ] for the polarization - dependent part . here @xmath24 with @xmath25 the quark mass . to describe the interaction with the target in we use the wilson line formalism in the @xmath26 light cone gauge of the projectile ( see [ ] for a detailed description of the formalism ) .
( our light cone 4-vector components are defined by @xmath27 . ) defining the fundamental representation wilson line @xmath28\ ] ] we can write the @xmath29-matrix operator for a fundamental - representation color dipole scattering on the target proton or nucleus by @xmath30.\ ] ] using this object we can write the interaction with the target @xmath14 as @xmath31 where the first two terms are given by the graphs in the first line of . our goal is to find whether the quark production cross section contributes to the stsa defined in .
clearly stsa should be given by the part of the cross section odd under the @xmath32 interchange .
stsa is usually thought of as arising from the @xmath33 term in the cross section , where @xmath34 is the spin of the incoming projectile and @xmath35 is its momentum : such term is @xmath0-odd .
analyzing we see that the @xmath32 transformation can be thought of as the @xmath36 coordinate interchange .
the @xmath0-odd stsa - generating part results from the part of the integrand in which is odd under the @xmath37 interchange : since the integrand is a product of @xmath13 and @xmath14 the @xmath0-odd contribution may arise in either of these terms . because our lowest - order @xmath13 is symmetric under @xmath36 , at this level of approximation
the stsa can only arise in the interaction with the target @xmath14 .
therefore we need to decompose into the @xmath36 symmetric and anti - symmetric parts , with the latter one giving rise to stsa when used in . before we proceed to anti - symmetrize @xmath14 ,
let us pause and point out that the decomposition into the light cone wave function squared and the interaction with the target is quite general and is valid up to corrections suppressed by powers of energy , which are usually neglected in the high energy qcd framework . the conclusion that the @xmath0-odd stsa - generating term may arise either in the wave function squared or in the interaction with the target is also quite general .
an example of the wave function squared giving the stsa - generating contribution is the initial - state sivers effect [ ] , though , as was illustrated in [ ] , and as is also seen in , the wave function squared also includes final state splittings and interactions . below we will construct the first ever example of the @xmath0-odd stsa - generating term generated in the interaction with the target . to anti - symmetrize @xmath14 under the @xmath38 interchange we first need to decompose each dipole @xmath29-matrix into the even and odd pieces under the exchange of its transverse coordinates , which corresponds to the @xmath39-parity operation exchanging the quark and the anti - quark [ ] : [ eq - pomeron and odderon ] @xmath40 the @xmath39- and @xmath0-even part @xmath41 is the `` standard '' dipole amplitude giving the total unpolarized dis cross section in dis .
its expression in terms of wilson lines allows to include glauber - mueller multiple rescatterings [ ] along with the nonlinear small-@xmath42 bk / jimwlk evolution in the leading-@xmath43 approximation .
the @xmath39- and @xmath0-odd _ odderon _ amplitude @xmath44 [ ] starts out with the triple gluon exchange at the lowest order [ ] , but then can also be enhanced by the multiple rescatterings and small-@xmath42 evolution [ ] . using the decomposition we can split @xmath14 into the @xmath0- and @xmath39-even @xmath37 symmetric part and the @xmath0- and @xmath39-odd @xmath37 anti - symmetric part : [ eq - symmetrized interaction ] @xmath45 the spin - dependent and spin - averaged cross sections @xmath46 and @xmath47 for quark production from read [ eq - observables ] @xmath48 where the wave functions squared are given by eqs . , , and the interactions are given by eqs . .
are the main results of this work : when substituted into they give the single - transverse spin asymmetry @xmath49 generated in quark production by the @xmath39- and @xmath0-odd interactions with the target .
the expressions in eqs .
are hard to evaluate in general .
to understand the main properties of the stsa contribution resulting from our mechanism , we used the amplitudes @xmath50 and @xmath51 evaluated in the glauber - mueller [ ] multiple - rescattering approximation and determined the corresponding stsa in the large-@xmath52 limit after making several crude approximations described in [ ] .
the results are summarized below .
first of all , in the large-@xmath3 limit the stsa due to our mechanism is ( assuming that the unpolarized cross section @xmath47 is dominated by gluon production ) @xmath53 which is a steeply - falling function of @xmath3 .
this indicates that in the standard factorization framework our stsa generating mechanism originates in some higher - twist operator . + the @xmath3-dependence of @xmath49 is illustrated over a broader momentum range in .
note that the asymmetry is a non - monotonic function of @xmath3 which even changes sign ( it has zeros / nodes ) .
this seems to be in qualitative agreement with some of the experimental data on stsa .
the maximum of @xmath49 in the saturation framework is correlated with the saturation scale .
another important feature is that the asymmetry in increases with the increasing momentum fraction @xmath16 ( an analogue of the feynman-@xmath42 variable ) for most values of @xmath16 : this is also in qualitative agreement with experiment .
an important feature of our stsa is that it is proportional to @xmath54 ^ 2\ ] ] with @xmath55 the nuclear profile function .
since the variation of @xmath55 is largest in the peripheral collisions , our stsa seems to be dominated by large impact parameter scattering and is sensitive to the edges of the target .
this explains the difference between the two panels in , which are different in the infrared cutoff used to limit the large-@xmath56 integration . a full analysis of the obtained formula without crude approximations made in [ ]
should determine whether the saturation scale is large enough in the peripheral collisions dominating stsa for the perturbative approach to be valid .
the dependence of our stsa on the size of the target for @xmath57 is @xmath58 if @xmath59 .
this is a very steep falloff of stsa with the atomic number of the nuclear target , which is also illustrated in , where @xmath49 is plotted for several target radii .
we thus predict that , in our formalism , stsa should be much smaller in the polarized proton
nucleus collisions than in the polarized proton proton collisions .
the mechanism we employed allows one to calculate the stsa of prompt photons as well .
the calculation yields [ ] @xmath60\end{aligned}\ ] ] where @xmath61 is given by with the @xmath62 replacement .
this expression is zero since @xmath63 due to the fact that the odderon amplitude is an anti - symmetric function of its transverse coordinate arguments , @xmath64 we thus have an exact result that in our mechanism the photon stsa is zero , @xmath65 at least in the order of approximation considered here .
in the future it will be important to explore other ways of generating stsa in the saturation / cgc formalism .
above we studied stsa originating in the interaction with the unpolarized target .
however , as we mentioned , it may also originate in the wave function squared .
this would be the sivers mechanism as studied in [ ] .
the analogue of this effect is illustrated in , where the stsa should be obtained by the interference between the diagrams in figs .
[ sivers ] and [ qtoqgampl ] .
the relative phase would arise due to the extra rescattering with the polarized projectile , and is depicted by the cut giving the imaginary part .
( indeed shows only one of the relevant diagrams . )
the analysis of this contribution is left for future work [ ] .
this research is sponsored in part by the u.s .
department of energy under grant no .
de - sc0004286 .
j. jalilian - marian and y. v. kovchegov , _ saturation physics and deuteron gold collisions at rhic _ , _ prog .
* 56 * ( 2006 ) 104231 , [ http://xxx.lanl.gov/abs/hep-ph/0505052 [ hep - ph/0505052 ] ] .
y. v. kovchegov and a. h. mueller , _ gluon production in current nucleus and nucleon nucleus collisions in a quasi - classical approximation _ , _ nucl .
phys . _ * b529 * ( 1998 ) 451479 , [ http://xxx.lanl.gov/abs/hep-ph/9802440[hep-ph/9802440 ] ] .
s. j. brodsky , d. s. hwang , and i. schmidt , _ final state interactions and single spin asymmetries in semi - inclusive deep inelastic scattering _ , _
phys.lett . _ * b530 * ( 2002 ) 99107 , [ http://xxx.lanl.gov/abs/hep-ph/0201296[hep-ph/0201296 ] ] .
j. c. collins , _ fragmentation of transversely polarized quarks probed in transverse momentum distributions _ , _ nucl.phys . _ * b396 * ( 1993 ) 161182 , [ http://xxx.lanl.gov/abs/hep-ph/9208213 [ hep - ph/9208213 ] ] .
y. v. kovchegov , l. szymanowski , and s. wallon , _ perturbative odderon in the dipole model _ , _ phys.lett . _ * b586 * ( 2004 ) 267281 , [ http://xxx.lanl.gov/abs/hep-ph/0309281[hep-ph/0309281 ] ] .
dedicated to the memory of jan kwiecinski . | we present the first steps in an effort to incorporate the physics of transverse spin asymmetries into the saturation formalism of high energy qcd .
we consider a simple model in which a transversely polarized quark scatters on a proton or nuclear target .
using the light - cone perturbation theory the hadron production cross section can be written as a convolution of the light - cone wave function squared and the interaction with the target . to generate the single transverse spin asymmetry ( stsa ) either the wave function squared or the interaction with the target has to be @xmath0-odd . in this work
we use the lowest - order @xmath1 wave function squared , which is @xmath0-even , generating the stsa from the @xmath0-odd interaction with the target mediated by an odderon exchange .
we study the properties of the obtained stsa , some of which are in qualitative agreement with experiment : stsa increases with increasing projectile @xmath2 and is a non - monotonic function of the transverse momentum @xmath3 .
our mechanism predicts that the quark stsa in polarized proton
nucleus collisions should be much smaller than in polarized proton
proton collisions .
we also observe that the stsa for prompt photons due to our mechanism is zero within the accuracy of the approximation . |
among the quantitative , hard " sciences , astronomy has traditionally been scorned , with particular disdain reserved for cosmology .
no more . in the decade of the nineties the combination of an avalanche of high quality observational data and theoretical advances driven by enhaced computer ( and brain ) power , have succeeded in transforming cosmology to a precise science . in this introductory lecture to iau symposium 198 on the light elements and their evolution it is my intent to describe primordial nucleosynthesis in this precision era of cosmology and to highlight the challenges , along with some goals , for the new millennium . after a brief review of the important physics during the era of primordial nucleosynthesis in the standard , hot big bang cosmological model ( sbbn ) , i will present an overview of the predicted primordial abundances , emphasizing the generally very small theoretical uncertainties .
these will then be compared to the present best estimates ( including their uncertainties ) of the primordial abundances inferred from current observational data .
after assessing the consistency of sbbn , i will explore what sbbn has to offer to cosmology and to particle physics and , what cosmology may teach us about sbbn .
i will conclude with a summary of the key issues / problems confronting sbbn and with a wish list of topics i hope will be addressed during this meeting and beyond .
our story begins when the universe is a few tenths of a second old and the temperature of the cosmic background radiation has dropped to a few mev as the universe expanded and cooled from its denser , hotter infancy . at this time
( and earlier ) the density and average energy of colliding particles is so high that even the weak interactions occur sufficiently rapidly to establish equilibrium . in particular , at this stage all flavors of neutrinos ( @xmath4 ) are in thermal equilibrium with the cosmic background radiation ( cbr ) photons and with the copius electron - position pairs present ( @xmath5 ) .
however , as the universe ages beyond a few tenths of a second and the temperature drops below a few mev , these weak interactions become too slow to keep pace with the rapid expansion of the universe and the neutrinos decouple from the cbr .
the electron - type neutrinos continue to play a role in transforming neutrons into protons and , vice - versa ( @xmath6 , @xmath7 , @xmath8 ) . as the temperature continues to drop , less
massive protons are favored over the more massive neutrons and the @xmath9 ratio falls ( roughly as @xmath10 , where @xmath11 is the neutron
proton mass difference @xmath12 mev ) .
after the temperature drops below 800 kev or so , when the universe is a few seconds old , even these weak interactions become too slow to keep pace with the expansion and the neutron - to - proton ratio freezes out " ( in fact , the ratio continues to decrease , albeit very slowly ) .
all the while , neutrons and protons have been colliding , occasionally forming deuterons ( @xmath13 ) .
however , the deuterons find themselves bathed in a high density background of energetic cbr photons which quickly photodissociate them ( @xmath14 ) before they can find a proton or neutron and form the more tightly bound , less fragile , h or he nuclei . since , as we shall see , there are roughly nine to ten orders of magnitude more cbr photons than nucleons in the universe , the deuteron stepping - stone " to further nucleosynthesis is absent until the temperature drops sufficiently low so that even in the high - energy tail of the black - body spectrum there are too few photons to prevent the deuteron from acting as a catalyst for primordial nucleosynthesis .
this critical temperature , which is weakly ( logarithmically ) dependent on the nucleon abundance ( the nucleon - to - photon ratio @xmath15 ) , is roughly 80 kev .
now , at last , when the universe is a few minutes old , big bang nucleosynthesis finally commences .
however , the universe was a fatally flawed nuclear reactor , cooling and diluting rapidly as it aged .
when the universe is some 10 20 minutes old ( @xmath16 sec ) and the temperature has dropped below 30 kev or so , the coulomb barriers preventing nuclear reactions between charged nuclei and protons and among charged nuclei become insurmountable ( in the short amount of time available ) and primordial nucleosynthesis comes to an abrupt end . in this all too brief but shining era
there has been time to synthesize ( in abundances comparable to those observed or observable ) only the lightest nuclides : d , he , he , and li . in
standard " ( a homogeneous universe , expanding isotropically with the particle content of the standard model of particle physics in which there are three flavors of light ( @xmath17 mev ) or massless neutrinos ) big bang nucleosynthesis ( sbbn ) the abundances ( relative to protons @xmath18 hydrogen ) of these four nuclides are determined by only one free parameter , the present epoch nucleon - to - photon ratio @xmath15 ( @xmath19 ) .
once the deuterium photodissociation bottleneck is breached primordial nucleosynthesis begins in earnest , quickly burning d to h , he and he .
the higher the nucleon density , the faster d is destroyed . the same is true of h ( which , if it survives will decay to he ) and he .
thus , the primordial abundances of d and he are determined by the competition between the nuclear reaction rates and the universal expansion rate .
the former rate depends on the overall density of the reactants the nucleon density . since all densities decrease as the universe expands , it is convenient to quantify the nucleon density by specifying the _ ratio _ of the nucleon density to the photon density ( measured after @xmath20 annihilation which enhances the universe s photon budget ) @xmath15 . since observations of the cosmic background radiation ( cbr ) temperature ( t = 2.73 k )
determine the present density of cbr photons , a knowledge of @xmath15 is equivalent to a determination of the present mass density in nucleons ( baryons " @xmath21 b ) . in terms of the density parameter @xmath22 ( the ratio of the mass density to the critical mass density ) and the present value of the hubble parameter ( h@xmath23 km / s / mpc ) , @xmath24 .
as @xmath15 increases the surviving abundances of d and he decrease ; since the he nucleus is more tightly bound than the deuteron , the decrease of the he / h ratio with @xmath15 is less rapid than that of d / h .
in contrast to d and he , the primordial abundance of he is not reaction rate limited since the nuclear reactions building helium-4 are so rapid that virtually all neutrons available when bbn commences are incorporated into he . as a result the he abundance , conventionally presented as the mass fraction of all nucleons which are in he , y@xmath25 , is _
neutron limited_. since the neutron - to - proton ratio is determined by the competition between the ( charged - current ) weak interactions which mediate the transformation of neutrons into protons ( and , vice - versa ) and the universal expansion rate , is sensitive to the universal expansion rate at the time the @xmath9 ratio freezes " and when the deuterium photodissociation barrier disappears . since the universal expansion rate is controlled by the total energy density
, provides an important test of cosmology and of particle physics in the early universe ( steigman , schramm & gunn 1977 ) .
it should be noted that is not entirely insensitive to the nucleon density since the higher @xmath15 , the earlier the photodissociation barrier is overcome . at earlier times when the temperature is higher ,
fewer neutrons have been transformed into protons and are available for incorporation into he . as a result , increases logarithmically with @xmath15 .
there is no stable nucleus at mass-5 and this presents a gap in the road to the synthesis of nuclei heavier than he . in order to bridge
the gap nuclear reactions must occur among nuclei with two or more nucleons .
but , the abundances of d , h , and he are small and the coulomb barriers ( especially between he and he and between he and he ) suppress these reactions as the universe expands and cools . as a result , there is very little leakage " to nuclei beyond mass-4 ; as a corollary , virtually all the he formed , survives .
the only heavier nucleus produced primordially in an abundance comparable to that observed ( or , even , observable with current technology ) is li , whose bbn abundance is some 4 5 orders of magnitude smaller than that of d and he .
the absence of a stable nucleus at mass-8 provides another gap preventing the production of astrophysically interesting abundances of any heavier nuclei . as will become clear in our subsequent discussion , the interesting " range of @xmath15 is @xmath26 ( @xmath27 ) , so we focus our discussion here on values of @xmath15 in this range . in the current precision era of bbn
_ most _ of the nuclear reactions relevant to the synthesis of the light elements have been measured to reasonable accuracy at energies directly comparable to the thermal energies at the time of primordial nucleosynthesis ( see nollett , this volume ) . as a result ,
the theoretical uncertainties in the bbn - predicted abundances are generally quite small . for @xmath15 in the above range , the 1@xmath28 uncertainties in d / h and he / h vary from 8 10% .
since he is most sensitive to the very well measured weak interaction rates , the error in sbbn - predicted is very small ( 0.2 0.5% or ,
@xmath29 = 0.0005 0.0011 ) . in contrast ,
larger uncertainties , of order 12 21% , afflict the predicted primordial abundance of li . since this symposium devoted much discussion to li , and space - limitations here
prevent me from discussing all the light elements in detail , i will concentrate in the following on the two key light elements , deuterium and helium-4 . in figure 1
is shown the relation between the bbn - predicted abundances of d and he .
the band going from upper left to lower right represents the @xmath30 range of uncertainties in the primordial abundances ( ( d / h)@xmath25 and y@xmath25 ) .
low d / h ( high @xmath15 ) corresponds to high and high d / h ( low @xmath15 ) corresponds to low y@xmath25 .
this anticorrelation will be very important when we confront the predictions of sbbn with the observational data .
to test sbbn and fully exploit the opportunities it offers to constrain cosmology ( e.g. , the baryon density ) and particle physics ( e.g. , new particles with weak or weaker - than - weak interactions ) requires that observational data be used to pin down the primordial abundances of the light elements to precisions as good as ( or , better than ) those of the sbbn predictions . as we approach the new millennium
there is good news along with some bad news .
the good news is that new detectors on ever larger telescopes which cover the spectrum from radio to x - ray energies and beyond are providing very high quality data , leading to inferred abundances of high statistical accuracy .
furthermore , the abundances of the light elements are determined from observations which differ from element to element in the telescopes and techniques employed as well as in the astrophysical sites explored . as a result , insidious correlated errors between and among the various element abundances are unlikely to be a problem .
the good news is also responsible for the bad news .
since the statistical errors have become so small , systematic errors now tend to dominate the uncertainties in the derived primordial abundances .
as bob rood has said during this symposium , estimating systematic errors is an oxymoron .
when a potential source of systematic error is identified , observations can ( and should ) be designed to eliminate or bound its contribution to the error budget .
it is a pointless and potentially misleading exercise to estimate " the magnitude of unidentified systematic errors . in part to remind us that our precise abundance determinations may not be accurate , and in part to challenge our observational colleagues who have done such a magnificent job of reducing the statistical errors , i will try to focus on the potential sources of systematic uncertainty ( when i can identify them ) in the following overview of the current observational status . as j. linsky ( this volume )
has reminded us , the deuterium abundance in the local interstellar medium ( the local interstellar cloud : lic ) is known very accurately : ( d / h)@xmath31 ( linsky 1998 ) .
since deuterium is only destroyed during the evolution of the galaxy ( epstein , lattimer & schramm 1976 ) , the lic abundance provides a lower bound to its primordial ( pre - galactic ) value .
this bound is strong enough to bound the nucleon density from above ( @xmath32 ; @xmath33 ) , ensuring that baryons can not
close " the universe ( @xmath34 ) , nor even dominate its present mass density ( @xmath35 ) .
thus , local observations of deuterium , combined with the _ assumption _ of the correctness of sbbn ( which we must test ) , already reaps great rewards : the mass - energy density of the universe must be dominated by unseen ( dark " ) non - baryonic matter . to go beyond ( in the quest for the primordial deuterium abundance )
we must look for observing targets which are less evolved than the lic .
the presolar nebula is one such site . from solar system observations of he reported by g. gloeckler ( this volume ) ,
it is possible to infer the presolar deuterium abundance ( geiss & reeves 1972 ; geiss & gloecker 1998 ) : ( d / h)@xmath36 . although marginally higher than the lic abundance , the larger errors prevent us from using this determination to improve on our previous bounds from the lic .
what this result does indicate is that there has been very little ( if any ) evolution in the d - abundance in the solar vicinity of the galaxy in the last 4.5 gyr .
this is consistent with a large class of galactic chemical evolution models discussed by m. tosi ( this volume ) which point to only a modest overall destruction of primordial deuterium by a factor of 2 3 ( tosi et al . 1998 ) .
if this theoretical estimate is combined with the lic abundance , we may estimate the primordial abundance : ( d / h)@xmath37 ( @xmath38 ) .
although possibly model dependent , this estimate is in remarkable agreement with the 2 3 determinations of d / h in high - redshift , low - metallicity ( hence very nearly primordial ) ly-@xmath39 absorbers illuminated by background qsos described by d. tytler and s. levshakov ( this volume ) . the data and analysis of burles & tytler ( 1998a , b : bt ) suggests that ( d / h)@xmath40 ( @xmath38 ) .
notice that the 1@xmath28 uncertainty in the observationally determined primordial abundance , @xmath41% , is impedance - matched to the @xmath41% sbbn theoretical uncertainty cited earlier .
however , lest we risk dislocating a shoulder while patting ourselves on the back at the triumph of such wonderful data , we should not ignore the claim ( webb et al .
1997 ; tytler et al . 1999 ) that the deuterium abundance in at least one ly-@xmath39 absorption system may be much higher .
this is a reminder that while any determination of the deuterium abundance anywhere in the universe ( lic , solar system , ly-@xmath39 absorbers , etc . )
provides a _ lower _ bound to primordial deuterium , finding an upper bound is more problematic .
indeed , in some absorbing systems it may be impossible to distinguish d - absorption from that due to hydrogen in an interloping , low column density , wrong - velocity " system .
thus , the deuterium abundance inferred from absorption - line data may only provide an _ upper _ bound to the true deuterium abundance .
since the low - z , high - z qso absorbing systems hold the greatest promise of revealing for us nearly unevolved , nearly primordial material , we look forward to the time when we can use the _ distribution _ of d / h values from more than a handful of such systems to eliminate statistically the uninvited contribution to the inferred primordial deuterium abundance from such interlopers . keeping this in mind , in the following i will , nevertheless , use the bt determination when confronting theory with data .
in contrast to deuterium whose primordial abundance only decreases as pristine gas is incorporated into stars , stars burn hydrogen to helium . as a result ,
the he observed anywhere in the universe is an unknown mixture of primordial and stellar - produced helium .
it has long been appreciated that to minimize the uncertain correction due to the debris of stellar evolution , it is best to concentrate on he abundance determinations in the lowest - metallicity regions available .
these are the low - z , extragalactic regions which have been discussed by k. olive , t. thuan , and s. m. viegas at this symposium ( this volume ) .
the reader is urged to consult their papers for details ; here i will merely summarize my view of the current status of the determination of the primordial he mass fraction y@xmath25 .
several years ago olive & steigman ( 1995 : os ) gathered together the data from the literature ( dominated by the data assembled by pagel et al .
more recently olive , skillman & steigman ( 1997 : oss ) supplemented this with newer data ( some of it , unfortunately , still unpublished ) . using a variety of approaches such as the regression of y on the oxygen and/or nitrogen abundances and the weighted means of y in the lowest metal - abundance regions , oss concluded that = 0.234 @xmath42 0.003 ( note that , in contrast to the published ( oss ) result , this value is obtained when the nw region of izw18 , suspected of being contaminated by underlying stellar absorption , is excluded from the fit , and the newer data of izotov , thuan and collaborators is not included ) .
izotov , thuan and their collaborators ( izotov , thuan , & lipovetsky 1994 , 1997 ; izotov & thuan 1998(it ) ; thuan , this volume ) have been systematically observing a mostly independent set of regions .
although , as with the data employed in the os and oss studies , they ignore the ionization correction ( @xmath43 ) , they take special care with the correction for collisional excitation .
it ( also thuan , this volume ) find y@xmath44 0.002 .
comparing the it and oss estimates of we find that difference between the two estimates far exceeds the statistical errors , suggesting systematic differences in the acquisition and/or analysis of the data samples . in a recent discussion which attempted to account for these unidentified systematic differences , olive , steigman & walker ( 1999 : osw )
combined the 2@xmath28 ranges for each determination to conclude : = 0.238 @xmath42 0.005 ; at the 2@xmath28 level , @xmath45 .
note , that this is also the 2@xmath28 upper bound to the it data alone .
since , as we shall see shortly , it is the upper bound which is crucial to testing the consistency of sbbn , in the following we shall adopt the it value ( and error estimate ) for the primordial abundance of he .
recently , viegas , gruenwald & steigman ( 1999 : vgs ; see viegas & gruenwald , this volume ) have emphasized the importance of the ionization correction which has heretofore been ignored .
vgs suggest that the it helium abundance ( y@xmath25 ) should be reduced by 0.003 to account for unseen neutral hydrogen in regions where the helium is still ionized in regions ionized by young , hot , metal - poor stars . in subsequent comparisons i shall explore the implications of adopting y@xmath46 0.002 . the cosmic history of the two other light nuclides produced in astrophysically interesting abundances during sbbn , he and li , is considerably more complex than that of d or he , which limits their utility as probes of the consistency of sbbn .
he is destroyed in the hotter interiors of all stars , but some he does survive in the cooler , outer layers . for lower mass stars this he survival layer increases and , indeed ,
newly synthesized he is produced by incomplete hydrogen burning .
the competition between destruction , survival , and synthesis complicates the galactic history of the he abundance .
nonetheless , since any deuterium incorporated into stars is first burned to he , the apparent lack of enhanced he ( see bania & rood , this volume ) argues against a very large pre - galactic abundance of deuterium ( steigman & tosi 1995 ) . for further discussion of the evolution of he see bania & rood ( this volume ) . as with he , any li incorporated into stars
is quickly burned away .
however , fusion and spallation reactions between cosmic ray nuclei and those in the interstellar medium are a potent source of li ( as well as of li , @xmath3be , @xmath47b , and @xmath48b ) .
it is also likely that there are stellar sources of li as indicated by the sample of lithium - rich red giants ( v. smith , this volume ) . since the abundance of lithium in the solar system and in the interstellar medium ( here and now " )
greatly exceeds that in the very metal - poor halo stars ( t. beers & s. ryan , this volume ) , the latter likely provide the closest approach to a nearly primordial sample .
since a significant fraction of this symposium is devoted to lithium , i will defer here to those other discussions except to comment that , within the theoretical and observational uncertainties , the primordial abundances inferred from the observational data are consistent with sbbn constrained by the confrontation with d and he .
although sbbn does lead to the prediction of the abundances of d , he , he , and li , the currently best - constrained primordial abundances are those of deuterium and helium-4 which we are concentrating on in this status report . for each value of @xmath15
, sbbn predicts a pair of ( d / h)@xmath25 and values .
therefore , in sbbn there is a unique connection between ( d / h)@xmath25 and which , allowing for the theoretical uncertainties discussed above , is shown as the band ( solid lines ) in figure 1 going from the upper left to the lower right ( 2@xmath28 uncertainties ) .
note that high - helium correlates with low - deuterium and , vice - versa .
also shown as the dotted ellipse in figure 1 is the contour of the ( independent ) 2@xmath28 uncertainties in the bt deuterium abundance and the it helium-4 mass fraction .
although the overlap between theory and data is not complete , figure 1 shows that , at the @xmath38 level , the predictions of sbbn are consistent with current observational data .
this is a dramatic success for the standard hot , big bang cosmological model .
of course it is not at all surprising that some value of @xmath15 may be found to provide consistency with the inferred primordial deuterium abundance .
but there was no guarantee at all that the helium-4 abundance corresponding to this choice would bear any relation to its inferred primordial value .
consistency with the bt d - abundance limits the nucleon abundance to the range ( 2@xmath28 ) @xmath49 or , @xmath50 .
for @xmath15 in this range there is consistency , within the theoretical and observational uncertainties , between the sbbn - predicted and observationally inferred primordial abundances of he and li as well .
four for the price of one !
there is , of course , one more test and opportunity offered by this result .
this sbbn - inferred nucleon abundance must also be consistent with present epoch estimates of the baryon density .
indeed , the sbbn - determined value of @xmath22 is _ larger _ than estimates ( persic & salucci 1992 ) of the luminous " matter in the universe suggesting that the majority of baryons are dark " .
this is good ( @xmath51 ) ; the opposite would have been a disaster .
this early - universe estimate of the baryon density is in good agreement with that inferred from the x - ray cluster baryon fraction ( steigman , hata & felten 1999 ) and with the independent estimate from the ly-@xmath39 forest ( weinberg et al .
1997 ) discussed below .
x - ray clusters likely provide a fair " sample of the universal baryon _ fraction _
@xmath52 ( white et al . 1993 ; steigman & felten 1995 ; evrard , metzler , & navarro 1996 ) which , when combined with the sbbn - inferred baryon density @xmath22 , leads to a clean " prediction , independent of detailed cosmological models , of the overall matter density @xmath53 .
if the results presented here are combined with the determination of @xmath52 from evrard ( 1997 ) , and with a hubble parameter @xmath54 ( mould et al .
1999 ) , we predict @xmath55 , in excellent agreement with several other recent , independent determinations . for example , a lower bound to the cosmic baryon density follows from the requirement that the high - redshift intergalactic medium contain enough neutral hydrogen to produce the ly-@xmath39 absorption observed in quasar spectra . according to weinberg et al .
( 1997 ) , depending on estimates of the quasar uv background intensity , this lower bound corresponds to @xmath56 , in excellent agreement with the sbbn prediction based on the bt deuterium determination . note that this lower bound from the ly-@xmath39 absorption forbids ( in the context of sbbn ) the primordial deuterium abundance to be any larger than @xmath57 , largely excluding the one surviving claim of high d ( webb et al . 1997 ) . indeed , if the sbbn results are combined with the magnitude - redshift data from surveys of high - redshift type ia supernovae ( garnavich et al .
1998 ; perlmutter et al . 1999 ) which bound a linear combination of @xmath53 and the cosmological constant @xmath58h@xmath59 , we may also constrain the cosmological constant ( @xmath60 ) , the curvature ( @xmath61 ) , and the deceleration parameter ( @xmath62 ) . as we have just seen , the sbbn - determined baryon density is consistent with that determined or constrained by observations of the universe during its present or recent evolution .
we may turn the argument around and ask what baryon density is suggested by non - bbn contraints , and then compare the light element abundances which correspond to that density with those inferred from the observational data . as an exercise of this sort , suppose ( for reasons of naturalness " or inflation ) that the universe is flat " : @xmath63 .
when combined with the sn ia magnitude - redshift data ( perlmutter et al .
1999 ) , this suggests that @xmath64 ( and @xmath65 ) .
now , if this mass density estimate ( @xmath53 ) is combined with with the x - ray determined cluster baryon fraction @xmath52 ( evrard 1997 ; steigman , hata & felten 1999 ) , the resulting nucleon abundance is @xmath66 .
although the uncertainty is large , it is reassuring that this non - bbn estimate has significant overlap with our sbbn estimate .
indeed , for the baryon density in this range sbbn predicts : ( d / h)@xmath67 and y@xmath68 .
the expansion rate of the early universe is controlled by the density of the relativistic particles present . in the standard model at the time of bbn these are : photons , electron - positron pairs ( when t @xmath69 ) and three flavors " of neutrinos ( @xmath70 , @xmath71 , @xmath72 ) which , if light " ( @xmath73 mev ) , are relativistic at bbn even if one or more of them may contribute to the present density of non - relativistic ( hot " ) dark matter . if
new " particles were to contribute to the energy density at bbn , the increase in the density would result in an increase in the universal expansion rate , leaving less time for neutrons to transform into protons
. the higher @xmath9 ratio at bbn would result in the production of more primordial he ( steigman , schramm & gunn 1977 ) .
it is convenient ( and conventional ) to characterize such additional contributions to the energy density by comparing their effects to that of an additional flavor " of ( light ) neutrino : @xmath74 . for @xmath75 small , @xmath76y @xmath77 .
notice in figure 1 that the predicted he abundance is a little high for perfect overlap with the observations . if @xmath75 were @xmath78 , ( n@xmath79 ) the overlap would improve ( e.g. , hata et al .
1995 ) , while if @xmath80 , the overlap would be reduced until it disappeared .
this is illustrated in figure 2 which shows the y versus d / h bbn band that would result if @xmath81 ( i.e. , @xmath82 , in contrast to the sbbn value of 3.0 ) .
notice that due to the faster expansion , more deuterium survives being burnt away so that , for fixed @xmath15 , the d - abundance also increases ; however since d / h is a much more sensitive function of @xmath15 , this has a much smaller effect on the y versus d / h relation than does the increase in y.
the study of the early evolution of the universe and , in particular primordial nucleosynthesis , has truly entered the precision era of cosmology .
precise abundances of the light nuclides are predicted and inferred from observations and the two are apparently in excellent agreement . as pleased as we may be at this success , it behooves us to avoid the temptation to rest on our laurels and to test this consistency ever more carefully . to this end
, it does nt take much contemplation to identify several clouds looming on the horizon .
what follows is my personal list of some problems / issues i would like to see addressed at this symposium and beyond .
first consider deuterium .
on the one hand , any determination of the d / h ratio anywhere , anytime provides a _ lower _ bound to the primordial abundance . on the other hand , since
wrong " velocity hydrogen can masquerade as deuterium , any observation of deuterium " is really an _ upper _ bound to its true abundance .
more data tracking the velocity structure of the absorbing features used to identify d and h and exploring variations in d / h in material with similar histories will be very valuable
. more data at high - redshift and low - metallicity will be very valuable .
after all , at present we are drawing profound conclusions on the basis of only two such systems .
much remains to be done concerning the primordial abundance of he .
for the most part , the regions from which the helium abundance is inferred have been modelled as homogeneous spheres or plane - parallel slabs .
a glance at the beautiful hst images of real regions reveals that they are anything but such idealizations .
what are the effects of temperature and/or density inhomogeneities , and how large may they be ?
what of underlying stellar absorption which , if present but neglected , would lead to an _
under_estimate of the helium abundance . and , what of the usually neglected ionization correction for neutral hydrogen and helium ( viegas , gruenwald & steigman 1999 ; see viegas & gruenwald , this volume ) ? considering this latter work , where models of regions ionized by realistic spectra of young star clusters were used in a reanalysis of the it data , a _ reduction _ in of order 0.003 was suggested .
a comparison with figure 1 shows that if were reduced by this amount , the overlap between theory and data would , in fact , disappear .
a half - dozen or so observations of deuterium in high - z , low - z systems along the lines - of - sight to distant quasars , with d / h determined in each ( on average ) to 10% or better .
with such a gift , i could determine @xmath15 to better than 4% , predict to @xmath83 , and constrain @xmath75 to an uncertainty less than @xmath42 0.05 .
i d be a very happy cosmologist indeed .
my second wish is for he measured to 3% accuracy ( or better ) in each of about a dozen low - metallicity , extragalactic regions , with care taken to address the several problems outlined above . with such data , could be fixed to better than the current level of @xmath84 , permitting he to be used as a baryometer ( @xmath85 ) .
much of what i know about this subject i have learned from my collaborators and i would be remiss if i failed to thank them for their contributions .
in particular , i wish to acknowledge r. gruenwald , k. olive , e. skillman , m. tosi , and s. m. viegas and , of course , my late friend dave schramm .
l. da silva , m. spite and the scientific and local organizing committees deserve great credit for their efficient organization of a very enjoyable and successful meeting . in part , this work is supported at the ohio state university by doe grant de ac0276er01545 . | the physics of the standard hot big bang cosmology ensures that the early universe was a primordial nuclear reactor , synthesizing the light nuclides ( d , he , he , and li ) in the first 20 minutes of its evolution . after an overview of nucleosynthesis in the standard model ( sbbn ) ,
the primordial abundance yields will be presented , followed by a status report ( intended to stimulate further discussion during this symposium ) on the progress along the road from observational data to inferred primordial abundances .
theory will be confronted with observations to assess the consistency of sbbn and to constrain cosmology and particle physics .
some of the issues / problems key to sbbn in the new millenium will be highlighted , along with a wish list to challenge theorists and observers alike .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in he@xmath0he he@xmath1he
li@xmath2li li@xmath3li h@xmath0h |
for millennia humans have observed the classical planets as unresolved points of light moving on the celestial sphere .
now we are on the verge of seeing _ extrasolar _ planets as unresolved , moving points of light .
we also are exploiting , or soon will exploit , alternative methods of imaging extrasolar planets , one of which we propose in this paper : sea - surface glints .
astronomical research and human exploration have a strong historical precedent .
the global oceanic explorations of captain james cook and others were in large part motivated by the possibility of measuring the scale of the solar system in physical units , e.g. meters , by geometrical triangulation of the silhouette of venus transiting the sun as viewed from nearly opposite sides of the earth during the transit of venus of 1769 ( maor 2000 ) .
cook s expedition required the significant resources of the greatest naval power on earth to sail for nearly 8 months from england to tahiti . that is , he journeyed as far as was humanly possible in the 18th century in order to attempt to answer a scientific question as old as humankind , `` how far is the sun , and hence by elementary geometry , how far are the other planets ? ''
although cook and others were blessed with cloud - free skies on the day of the transit , the scientific objective of his expedition failed due to turbulence in the earth s atmosphere . from the long view of history
, we do not recall such expeditions costs in lives or gold , nor does it matter much that they failed in their scientific objectives .
instead , we revere such expeditions for their power to inspire humans to collectively seek great accomplishments for the good of all humankind . in this paper , we outline a similar scientific quest to understand humankind s place in the cosmos : a historically significant and achievable scientific goal for nasa would be to identify _ oceanic _ extrasolar planets .
this goal is simpler than , and also related to , the goal of former nasa administrator dan goldin to image the surfaces of extrasolar planets .
if we were to attempt such imagery by diffraction - limited optics , the minuscule angular size of an extrasolar planet would require truly enormous separations between the optical components .
however , an indirect imaging technique such as that presented in section [ sec : glints ] may permit mapping the surfaces of extrasolar planets with telescopes that are well within the capabilities of current human technologies .
in section [ sec : context ] i provide some context to the lunar exploration initiative and some digressions on possible impacts it may have on society .
we need not spatially resolve an extrasolar planet in order to determine if it has an atmosphere and an ocean like those of earth .
instead , we propose to exploit the linear polarization generated by rayleigh scattering in the planet s atmosphere and specular reflection ( glint ) from its ocean to study earth - like extrasolar planets . in principle
we can map the extrasolar planet s continental boundaries by observing the glint from its oceans periodically varying as the rotation of the planet alternately places continents or water at the location on the sphere at which light from the star can be reflected specularly to earth .
the concepts in this section have been described by mccullough ( 2007 ) and independently by others .
seager et al .
( 2000 ) , saar & seager ( 2003 ) , hough & lucas ( 2003 ) , stam , hovenier , & waters ( 2004 ) , and stam & hovenier ( 2005 ) have examined the rayleigh - scattered light of a hot jupiter .
williams & gaidos ( 2004 ) and gaidos et al . ( 2006 ) examined the unpolarized variability of the sea - surface glint from an earth - like extrasolar planet .
stam & hovenier ( 2006 ) independently examined the observability of the polarized signatures of an earth - like extrasolar planet , including rayleigh scattering and sea - surface glint .
kuchner ( 2003 ) and lger et al .
( 2004 ) have proposed that `` ocean planets '' may form from ice planets that migrate inward and melt ; the surfaces of these planets would be liquid water exclusively , i.e. no continents , and it remains to be seen whether such planets would be entirely obscured by a thick steam atmosphere , or instead might have clear atmospheres like earth s .
specularly reflected light , or glint , from an ocean surface may provide a useful observational tool for studying extrasolar terrestrial planets .
an interesting parallel exists between using glints to image the oceans of extrasolar planets and a similar technique to image within the earth s turbid ocean ( e.g. moore et al .
2000 ) . in the latter ,
one aims an underwater camera toward the sea floor while illuminating the camera s field of view with a laser beam .
laser light scattered by the ocean water creates a haze of light visible by the camera , but the laser light that reaches the sea floor creates a well - defined spot that is also detected by the camera . by scanning the laser across the sea floor and simultaneously recording the location and brightness of the peak of the image ,
the light scattered by the turbid water is suppressed and detection of objects on the sea floor is enhanced . in the proposed technique for imaging extrasolar planets ,
the glint acts like the localized spot of the laser beam , and the rotation of the planet under the glint serves much the same purpose as the scanning of the laser beam .
the polarization of the glint allows one to isolate it from the less - polarized reflection from other regions on the surface , and from the nearly - unpolarized star light scattered in the telescope optics . in the underwater technique ,
the monochromatic color of the laser s light can be used to increase its contrast over any ambient light .
analogously , the two primary mechanisms of polarization , the glint and atmospheric rayleigh scattering , can be differentiated with color , since the former is nearly achromatic whereas the latter depends strongly on wavelength .
detection of sea - surface glints would differentiate ocean - bearing terrestrial planets , i.e. those similar to earth , from other terrestrial extrasolar planets .
the brightness and degree of polarization of both sea - surface glints and atmospheric rayleigh scattering are strong functions of the phase angle of the extrasolar planet ( see mccullough 2007 ) .
the difference of the two orthogonal linearly polarized reflectances may be an important observational signature of rayleigh scattering or glint .
the difference attributable to rayleigh scattering peaks near quadrature , i.e. near maximum elongation of a circular orbit .
the difference attributable to glint peaks in crescent phase , and in crescent phase the total ( unpolarized ) reflectance of the glint also is maximized .
the reader intrigued by this short summary will find additional detail in mccullough ( 2007 ) and references therein .
as quoted in his son s book ( dyson 2002 ) , freeman dyson wrote in may 1958 , _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we shall know what we go to mars for , only after we get there .... you might as well ask columbus why he wasted his time discovering america when he could have been improving the methods of spanish sheep - farming . it is lucky that the u.s . government like
queen isabella is willing to pay for the ships .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in the case of the earth s moon , we have been there already , and to me , the potential tangible benefits of lunar exploration are not as clear as the intangible benefit of inspiring people to reach collectively for grand accomplishments .
preparing for this conference has caused me to consider many things that i would nt have otherwise . in this section
i digress to consider ( in order ) the relative cost of nasa , some potential risks of the us not exploring the moon , a significant budgetary challenge to sustaining that initiative , some benefits and concomitant risks of the lunar - exploration initiative .
humans commonly redefine large quantities in appropriate units , such as the a.u . for solar system distances , or the parsec for stellar distances .
i need to do the same for the large sums of money associated with lunar exploration .
financially , nasa is approximately equivalent to a single large corporation . the median market capitalization of the 30 corporations in the dow jones industrial average is 108 b usd ( as of oct 31 , 2006 ) .
one such corporation , pfizer , a pharmaceutical company founded in 1849 and headquartered in nyc , in 2005 had annual revenue of 51 b usd and spent 7 b usd on r&d .
nasa by comparison was awarded 16 b usd in 2005 by the us congress .
the net worth of us households in 2000 was 50 t usd .
the aggregate value of corporate equities directly held was 9 t usd in 2000 and was 4 t usd in 2003 , so it declined approximately 2 t usd per year for three consecutive years .
that is 2000 b usd per year , or 1 b usd per hour of each and every working day for three years .
although it is hard for me as an individual to grasp these large sums , these comparisons may help put into context that nasa s lunar exploration initiative is expensive in one sense but not relative to the richness and power of our society , and especially not in comparison to those of the global society .
returning to the moon may make sense when one considers the potential risks or costs to the usa of _ not _ returning to the moon when other nations do .
each nation may have a self - interest in establishing a presence on the moon , much like antarctica , in order to assure that no single nation monopolizes it . as someone once said
when asked how much the us should spend on science , _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ we should spend exactly as much in each field as makes us first in that field . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ if a competitor were to establish a presence on the moon , the us might worry that it was missing out on something of which it was not aware . here , i am reminded of seward s folly , the purchase of alaska , which in fact was both a strategic windfall and an economic bargain . over the many years required to return to the moon
, there are large risks that could jeopardize the political will to continue the initiative .
the largest risk , in my estimation , is the us federal budget .
the aging of the `` baby boom '' demographic of the usa and western europe will soon greatly increase the rate of retirements .
beginning approximately in 2010 , which is 1945 ( the end of world war ii ) plus 65 years ( the nominal age of retirement ) , wage earners that had been paying income taxes and capitalizing equity markets will retire and begin to _ withdraw funds_. political recognition of this impending financial bust is evident ; for example , on jan 18 , 2007 , us federal reserve chairman ben bernanke testified on this specific topic before the senate budget committee , `` we are experiencing what seems likely to be the calm before the storm .... '' if somehow that bust does not materialize , or the lunar exploration initiative somehow is immune to its effects , i expect a very substantial , albeit largely intangible , benefit of lunar exploration may be to bring the peoples of the world closer together , to save ourselves from ourselves .
an excellent treatise on the latter topic was written at the time of the apollo program , `` the tragedy of the commons '' by garrett hardin ( 1968 ) .
hardin argues well that some problems do not have a _ technical _ solution .
he also points out that scientists , and often policy makers as well , often assume that a technical solution exists and fruitlessly seek one in their faith that one will eventually be found .
( consider global climate change . )
perhaps colonization of the moon will be a metaphor for solving earth s geopolitical problems .
for example , a quasi - sustainable presence on the moon would nt use fossil fuels ; it would use solar or nuclear power with a considerable emphasis on conservation . due to the large expense , in energy or dollars , of moving mass from the surface of earth to the surface of the moon , nuclear power , in any of the various forms of radioisotope thermoelectric generators , reactors , or explosives , may be convenient for lunar exploration for electrical power , transmuting elements _ in situ _ , or excavations .
however , any such convenience , i.e. of anything brought from earth , should be considered a negative compared to the long - term strategic benefit of `` living off the land ( regolith ) . '' for example , solar power is readily and abundantly available on the lunar surface , and a large , slow flywheel utilizing compacted regolith for mass could provide for the variable power demands of human habitation and/or store power through the lunar night away from the poles .
an alternative approach would be to bring from earth a high - speed , precision flywheel of relatively small mass or a chemical battery , but those are antithetical to the strategic benefit of utilizing lunar resources . from the opposite perspective , utilizing any water ice mined from lunar craters , for human consumption and/or rocket fuel , could be myopic exploitation and destruction of an important scientific resource .
the benefits of knowledge always have had concomitant negative consequences : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the idea that curiosity leads to disaster has an ancient pedigree .
pandora opened the gods box and let loose all the evils of the world ; the descendants of noah built the tower of babel to reach heaven , but god scattered them and confounded their language ; icarus flew so close to the sun that his homemade wings melted , and he fell into the sea and drowned ; eve ate the apple of knowledge and was exiled from the garden of eden ; faust traded his soul for sorcery and spent eternity in hell .... we believe that curiosity is the beginning of knowledge and especially of science , but we know that the application of science has led to disaster .
finkbeiner ( 2006 ) . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ neils bohr , as attributed by rhodes ( 1986 ) , believed the knowledge of nuclear weapons would `` foreclose the nation - state , '' by which he meant that knowledge of nuclear weapons was so powerful that its proliferation would make the nation - state obsolete .
today , one might imagine any number of pandora s boxes equally well in that role , but none yet that have been opened ( at least not for long ) outside governmental control .
it seems to me an inevitability akin to the second law of thermodynamics that diffusion of knowledge will occur , whether or not that diffusion is good for humankind .
technology increasingly amplifies the power of an individual , or a small group of individuals , for good or for evil . as archimedes said of the lever ,
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ give me a place to stand on , and i will move the earth . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ here i suggest the lesser task of moving an asteroid , as an example of something that seems entirely fanciful now , but which may not be so in decades hence .
consider again the example given in the introduction , of captain cook s voyage : at that time , it required government sponsorship , whereas today s transportation and technical infrastructure make replicating the feat entirely within the capacity of a private individual .
astronomers have suggested that they may give early warning of an asteroid on a collision course with earth .
technologies useful to measuring the orbit of such an asteroid with precision sufficient to enable the prediction of an impending collision , and especially those technologies useful to perturbing it to prevent a collision , are vaguely understood and only potentially available to powerful nations today .
mostly , those nations lack a specific motivation to act .
however , in some number of decades , what would have required the concerted effort of a nation or nations might be accomplished by a group of individuals , but instead of preventing a collision , that group could attempt clandestinely to create a collision from what naturally would have been a near miss . even a credible threat to do so would be influential .
presumably it is more difficult to turn a near miss into a collision than vice versa , and today we may take solace in our confidence that such a concept is entirely impractical .
my purpose is to illustrate the duality of technology s amplification of power to individuals with a novel example potentially relevant to the lunar exploration initiative and related technologies .
for those that reasonably consider the proposition patently absurd , i recite with intentional irony margaret mead , _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ never doubt that a small group of thoughtful , committed citizens can change the world . indeed , it is the only thing that ever has . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
an indirect method of imaging the oceans and continental boundaries of extrasolar planets is outlined .
results from simulations of earth - like extrasolar planets ( mccullough 2007 ) are presented in section [ sec : glints ] and demonstrate that the difference of fluxes in two orthogonal linear polarizations is modulated by the planet s rotation , as it alternately places continents or water at the location on the sphere at which light from the star can be reflected specularly to earth .
section [ sec : context ] digresses into the lunar exploration initiative s potential impacts on society and vice versa .
gaidos , e. , moskovitz , n , & williams , d. m. 2006 , iau colloq . 200 : direct imaging of exoplanets : science & techniques hardin , g. 1968 , science , 162 , 1243 hough , j. h. , & lucas , p. w. 2003 , esa sp-539 : earths : darwin / tpf and the search for extrasolar terrestrial planets , 11 kuchner , m. j. 2003 , apj l , 596 , l105 lger , a. , et al . 2004 ,
icarus , 169 , 499 mccullough , p. r. 2007 , apj , submitted , also arxiv astrophysics e - prints , arxiv : astro - ph/0610518 moore , k.d . , jaffe , j. s. , and ochoa , b. l. 2000 , journal of atmospheric and oceanic tech .
17 , no . 8 , pp . 1106 - 1117 , also http://jaffeweb.ucsd.edu/pages/instruments/lls/lls.html rhodes , r. 1986 , the making of the atomic bomb , simon & schuster , new york , p. 783 .
seager , s. , whitney , b. a. , & sasselov , d. d. 2000 , apj , 540 , 504 stam , d. m. , de rooij , w. a. , cornet , g. , & hovenier , j. w. 2006 , a&a , 452 , 669 stam , d. m. , & hovenier , j. w. 2005 , a&a , 444 , 275 stam , d. m. & hovenier , j. w. 2006 , iau colloq . 200 : direct imaging of exoplanets : science & techniques stam , d. m. , hovenier , j. w. , & waters , l. b. f. m. 2004 , a&a , 428 , 663 williams , d. m. , & gaidos , e. 2004 , bulletin of the american astronomical society , 36 , 1173 yardeni , e. 2004 , consumer handbook ( with baby boom charts ) , prudential equity group , llc , dated aug 13 , 2004 . | ambitious studies of earth - like extrasolar planets are outlined in the context of an exploration initiative for a return to the earth s moon .
two mechanism for linearly polarizing light reflected from earth - like planets are discussed : 1 ) rayleigh - scattering from a planet s clear atmosphere , and 2 ) specular reflection from a planet s ocean .
both have physically simple and predictable polarized phase functions .
the exoplanetary diurnal variation of the polarized light reflected from a ocean but not from a land surface has the potential to enable reconstruction of the continental boundaries on an earth - like extrasolar planet .
digressions on the lunar exploration initiative also are presented . |
in its beginnings , string theory was expected to be a possible model for strong interactions .
but it was eclipsed by the greatest qcd outcomes , making it to be forgotten as fundamental theory for such interactions .
nowadays , with the development of the so - called gravity / gauge duality , string theory recovers its original motivation becoming a tool to explore the strong interactions at their non - perturbative region even at extreme scenarios with finite temperature and chemical potential .
these scenarios are observed at nuclei accelerators and stars center .
duality itself suggests that gravitational dual of qcd should exist .
the real problem is how to find it . in this direction ,
the most successful approximation to gravity / gauge duality is the ads / cft correspondence by j. maldacena @xcite , which has been used as a tool to describe non - perturbative hadronic phenomena , specially the quark gluon plasma observed in heavy ion collisions .
properties such as the entropy density , bulk viscosity , jet quenching , photon / dilepton emission rate were well described holographically in the zero chemical limit by top / down approaches as the dp / dq system @xcite or the sakai sugimoto model @xcite
. also the bottom / up realizations as the improved holographic qcd model @xcite have described the qgp system .
recently , @xcite have discused the shear viscosity and photon emission in the qgp in the latter model . in the ads
/ qcd realm , some approximations have been made in the references @xcite where they calculated the spectral density for photons in a specific gravitational setup using the soft wall model .
also @xcite has calculations for the hard wall and the soft wall model for finite chemical potential modeled by the probe d - brane method , in which the dbi action is minimized in order to obtain the embedding profile generated by a u@xmath0 static field .
a good review of the holographic approximation to the qgp phenomenology is @xcite .
this paper is organized as follows : in section 1 we discuss the photon emission rate and the conductivity from the quantum field theory side . in section 2 , we set the general background for both hard wall and soft wall models . in section 4
it is discussed how we calculate the trace of the spectral function analytically for each ads / qcd model . in section 5 ,
we consider the numerical results obtained for the photon emission rate and the conductivity .
finally in section 6 we enunciate the conclusions for our work .
it is well known that any thermal medium composed of charged particles will radiate photons .
this photon spectrum can be used as a probe to study the characteristics of such thermal medium , since it depends on its specific properties . in the case of
the qgp produced in heavy ion collisions , the thermal medium is optically transparent due to the small spatial extension and the smallness of electromagnetic coupling .
therefore , any emitted photon escapes from the plasma ball without any further interaction with the medium . in such a situation
the spectrum can be considered as _ almost planckian _ , but it is still useful to obtain information about the plasma in the emission point . at this point
a remark must be done : depending on the momentum @xmath1 of the emitted photons the dynamics can be split in specific regimes defined by three characteristic length scales , the hard , soft and ultra - soft scales @xcite .
the hard scale is defined when the limit @xmath2 is reached .
contributions of excitations in this scale dominate the bulk thermodynamic properties such as total momentum density and total energy of the plasma , likewise conserved charge susceptibilities .
these excitations also control the transport properties in the colored medium .
the second scale , soft scale , is reached when @xmath3 , where @xmath4 is the gauge theory coupling . in this scenario
the landau damping and also the debye screening , due principally to the coherent interaction of the particles in the thermal colored medium , become important .
these effects are treated with the hard - thermal - loop perturbation theory @xcite .
the third scale is the ultrasoft or non - perturbative : @xmath5 . in this scale , low frequency and long wave - length dynamics gives the hydrodynamical regime .
photons produced in this regime carry information about properties such as the electrical conductivity .
soft and hard photons produced in more energetic scenarios tends to be suppressed due to the landau pomeranchuk midgal effect .
photon production in a colored medium was discussed in detail both from strong and weak coupling point of view in reference @xcite .
let us consider a qgp defined by a colored quantum system in equilibrium that can be described by a thermal field theory in the strongly coupling limit .
it will be assumed that the electromagnetic interaction between matter and photons is given by the minimal coupling of a photon field with the electromagnetic current , @xmath6 , with @xmath7 the electromagnetic coupling . at leading order ,
the photo - production is given by the expression @xcite @xmath8 where we have defined the photon 4-momentum as @xmath9 , @xmath10 denotes the minkowski metric and @xmath11 is the wightman function defined for electromagnetic currents as , @xmath12 notice that the expectation value in ( [ wightman ] ) is taken in the thermal equilibrium state and @xmath13 . in the thermal equilibrium limit the wightman correlator can be written in terms of the spectral density , @xmath14 with @xmath15 the bose - einstein distribution and @xmath16 the thermal retarded green function . at non zero temperature , lorentz invariance is explicitly broken leaving us a residual rotational symmetry that allows , together with the gauge invariance , to split the thermal green function into longitudinal and transversal parts . since the idea is to model real photon production ( implying @xmath17 ) , it is possible to ignore the longitudinal part and focus on the transversal one only .
then , the trace of the spectral function ( [ spectral2 ] ) is written as @xmath18 where @xmath19 is the transversal part of the spectral density . keeping in mind all these definitions and results ,
the photon emission rate is given by @xmath20 the electrical ac and dc conductivities of the medium are also determined by the trace of the spectral function . from the kubo formula the ac @xmath21 conductivity can be read from the spatial components of @xmath22 as @xmath23 from the spectral density in the limit of @xmath24 we can read the dc conductivity as @xmath25 then the problem of calculating the photon emission rate and the conductivity translates into the calculation of the trace of the spectral function holographically .
this procedure will be realized in the next sections .
geometric background we will focus on the calculation of the 2-point function using two of the ads / qcd models : the hard wall @xcite and the soft wall @xcite models .
these two approximations can be summarized into a single expression just by considering the behavior of a statical dilaton @xmath26 field . in both ads / qcd models
the background is considered dual to the colored medium properties , composed by massless quarks .
since we are considering the finite chemical potential case , it is possible to introduce a charge in the ads background through an einstein hilbert
maxwell action .
holographically , we will say that this charge is dual to the generator of the @xmath27 baryonic symmetry .
following these ideas the background action can be written as @xmath28 where @xmath29 , the cosmological constant is @xmath30 , with @xmath31 the ads radius , @xmath32 is the newton constant in five dimensions , and the five dimensional gauge coupling is defined as @xmath33 , with @xmath34 and @xmath35 the number of colors and flavors respectively . adding temperature to a given system holographically means the presence of a black hole , so it is possible to conclude that the minimal coupled hilbert
maxwell action ( [ bg_action ] ) leads to a the reissner
nordstrom ads black hole solution .
in other words , this rn ads bh solution can be considered as the gravitational dual for a strongly coupled plasma under finite chemical potential and finite temperature conditions @xcite . following these ideas ,
the background is written as : @xmath36,\ ] ] where the blackening factor @xmath37 is fixed by the 1-form potential @xmath38 .
if a static solution is chosen , @xmath39 , the @xmath40 factor can be written as @xmath41 where @xmath42 is the charge of the black hole and @xmath43 is the event horizon .
the temperature is defined using the hawking page relation @xcite @xmath44 in pure ads , i.e. without dilaton , the 1-fom potential is given by @xmath45 . in ads
/ qcd models the same form can not be granted due to the presence of the cutoff . in that case
it is possible to approximate the ads / qcd solution to the pure ads one in the limit @xmath46 @xcite .
holographically , the chemical potential @xmath47 related to the gauge theory in the boundary is defined in the limit @xmath46 where the 1-form static potential is @xmath48 .
holographic photons the qgp will be modeled using a 5-dimensional @xmath27 field @xmath49 coupled to the static dilaton in the ads / qcd standard form @xmath50 where @xmath51 is the coupling constant for photons in the qgp , that in general depends on temperature @xcite and @xmath52 . from the photon action it is possible to obtain the equations of motion for the vector field @xmath49 . to do this ,
the photons will be supposed to be moving in the @xmath53 euclidean direction , then the photon 4momentum can be fixed to be @xmath54 .
therefore , using the background ( [ metric_rn_ads ] ) the e.o.m . in fourier space
are written as @xmath55+\frac{e^{-\phi\left(z\right)}}{z}k_0 ^ 2\left[1-h\left(z\right)\right]a\left(k , z\right)=0,\ ] ] where the gauge fixing @xmath56 was implicitly used and @xmath57 is the transversal field component .
the longitudinal components , related to the dilepton production are fixed to be zero .
thermal green function the retarded green function @xmath16 can be constructed with the solutions of eqn .
( [ eqn_mov ] ) . in order to do that
, the first step is to evaluate the on - shell boundary photon action obtained from ( [ photons ] ) : @xmath58_{z\rightarrow 0}^{z = z_h}}.\ ] ] following the minkowskian prescription given in @xcite , the thermal green function can be obtained from the integrand in the boundary action , evaluated in the event horizon @xmath43 : @xmath59 with this expression , the trace of the spectral function ( [ spectral2 ] ) is given by @xmath60 the trace of the spectral function is calculated from the action ( [ photons ] ) using the static background ( [ metric_rn_ads ] ) . in the next sections
the discussion will be focused in each ads / qcd approximation , namely , hard wall and soft wall models .
hard wall model hard wall is defined by fixing the dilaton field to be zero , @xmath61 , together with the dirichlet condition on the transverse fields at some holographic coordinate @xmath62 taken as a hard cutoff , where @xmath63 is the qcd energy scale .
the physical behavior of the hard wall system is equivalent as the square well in quantum mechanics : the presence of the cutoff induces a discrete set of solutions that defines the eigenstates of the system . the background action in the hard model is then given by @xmath64,\ ] ] where @xmath65 is the usual step function that defines the ads slice and @xmath40 is the warp factor defined in the expression ( [ blackening ] ) .
the temperature in this case follows from ( [ temp ] ) . to obtain the charge in terms of the chemical potential @xmath47
it is necessary to solve the equation of motion for @xmath66 in the ads slice and then evaluate the hilbert
maxwell action to construct the gran canonical potential @xmath67 . in order to have a regular solution we will impose that @xmath68 . from the hilbert einstein
maxwell action evaluated in the hard wall background ( [ hwmbg ] ) it is possible write the equation of motion for the electrostatic potential @xmath69 as : @xmath70=0,\ ] ] subjected to the boundary conditions @xmath71 and @xmath72 .
the solution under these conditions is @xmath73 , and after evaluating the on shell hilbert
maxwell action , the grand canonical potential is written as @xcite : @xmath74 with @xmath75 a fixed volume in the usual euclidian space than can be fixed to be one and @xmath76 is a function of @xmath77 . from the classical thermodynamic relation @xmath78
we find the relation between the baryonic charge and the chemical potential .
then , we obtain : @xmath79 now that the background is fully characterized we can fix our attention on the photon action in the ads slice in order to obtain the trace of the spectral density @xmath80 .
the e.o.m . for the photon field in this hard wall scenario is @xmath81+\frac{k_0 ^ 2}{z\,h\left(z\right)}\left[1-h\left(z\right)\right]\,a\left(z\right)=0.\ ] ] to solve this equation , it is necessary to set the incoming boundary condition as the field @xmath82 approaches to the horizon . to do this
, the field @xmath82 will be written as @xmath83 , where the @xmath84 function is fixed to be 1 in @xmath76 . following these ideas we are able to write the trace of the spectral density :
@xmath85 where the coupling constant is defined as @xmath86 @xcite . this coupling measures the electric degrees of freedom in the plasma and scales with the temperature , since it enhances the photon production as a consequence of growing the plasma ball . in general
, equations like ( [ eomhw ] ) have no analytical solutions since their highly non - linear behavior , so we need to make use of numerical methods in build up the trace of the spectral densit .
in order to do this analysis we will define the adimensional frequency @xmath87 and baryonic density charge @xmath88 .
the results for the trace of the spectral function and the photon production are showed in the figure [ fig : figone ] .
soft wall model the static background in the soft wall is defined by the dilaton profile @xmath89 , where @xmath90 is an energy scale of the hadron spectra in a given regge trajectory . in this case , it is not univocally defined in terms of the temperature and baryonic density .
thus , as a first approximation , the values of @xmath91 will be taken small in order to use the relation @xmath92 given by @xcite .
the chemical potential can be obtained from the hilbert einstein
maxwell action ( [ bg_action ] ) in this background by virtue of the e.o.m . @xmath93 with the boundary conditions @xmath94 and
@xmath95 , the solution for the equation above is in the form @xmath96 in order to calculate the baryonic charge we will impose that the 1-form static potential in the limit @xmath46 should match the rn ads version of it @xcite . evaluating the on - shell hilbert
maxwell action and using the thermodynamic relation ( [ thermo ] ) we obtain the baryonic charge as a function of the chemical potential : @xmath97 once you have fully characterized the background , the next step is the construction of the spectral function starting from the e.o.m in this case @xmath98+\frac{k_0 ^ 2\,e^{-\kappa^2z^2}}{z\,h\left(z\right)}\left[1-h\left(z\right)\right]\,a\left(z\right)=0.\ ] ] imposing the ingoing boundary conditions as in the hard wall case we arrive to the trace of the spectral function given by the expression @xmath99 with the same @xmath100 coupling as in the hard wall is used here . using the trace of the spectral function it is possible to construct numerically the photon emission rate in terms of @xmath101 and @xmath102 .
the results in this approximation are plotted in the figure [ fig : figtwo ] .
in order to do the numerical analysis in frequency , the temperature is fixed while the baryonic charge is varying .
the results for hard wall and soft wall models are plotted in the figures [ fig : figone ] and [ fig : figtwo ] .
the hard wall model is a holographic bag in the sense that confinement is placed by a hard cutoff fixed by the temperature of the colored medium .
any other extra properties of the medium are hidden in the background .
for example , in dp / dq systems the quark mass is given by the embedding of the dq
branes into the dp black branes in the decoupling limit @xcite . in the case of ads
/ qcd models , this specific information can be added with extra fields in the hilbert
maxwell action .
one application of this idea can be seen in @xcite where the quark mass is introduced as a scale defined by a bulk mass in a static modified hard wall . in the case of the soft wall model ,
the cutoff is a decreasing function of the coordinate @xmath103 defined by a static dilaton .
the energy scale @xmath90 is fixed by the temperature of the medium .
in other applications of the soft wall model to hadronic spectra the energy scale is fixed by the lightest particle in the spectrum @xcite .
it is also possible to introduce massive quarks by adding extra fields into the background action .
this will be done in future works .
the trace of spectral density on both models behaves monotonically with @xmath102 for the explored values of @xmath101 .
the soft wall model is more sensitive to increasing of @xmath101 than the hard wall .
this is due to the presence of the dilaton field in the first one .
the scale @xmath90 is a function of @xmath77 and @xmath101 . for values of @xmath101 near to the upper limit , @xmath104 , the trace of the spectral function is expected to begin to oscillate because the background is unstable for those values of @xmath101 @xcite .
the results for @xmath105 are also in agreement with those obtained in @xcite , where the authors construct a different background in order to simulate the results for the d3/d7 system in the massless quark region and try to study how the trace of the spectral function behaves as a function of @xmath90 .
our background and the one showed in that reference differs by the transformation @xmath106 .
it is no surprising that these models have similar behaviors .
in fact , dp / dq systems and ads / qcd models are ads - like ( or conformal ads ) in the sense of @xcite .
the photon emission rate obtained for both models ( right panels in figures [ fig : figone ] and [ fig : figtwo ] ) are also consistent with the qgp phenomenology .
the emission occurs for small values of @xmath102 , where the ultrasoft behavior is expected @xcite .
when @xmath102 is increased the photon emission is damped , according to the landau damping effect and debye screening : photons are reabsorbed by the medium and are suppressed . for high values of @xmath102 ,
the emission goes to zero due to the reduction of bremsstrahlung and pair production , the so called landau pomeranchuk
migdal effect @xcite .
when the chemical potential is turned on , the chemical equilibrium is broken , implying that there is an excess in the quark ( or anti - quark ) population that makes the enhancement of the quark ( anti quark ) bremsstrahlung over their annihilation radiation , causing that the photon emission rate increases with the chemical potential for low energies @xcite .
for all the @xmath101 values tested , the photon emission rates have common asymptotic behavior with large @xmath102 , demonstrating the expected universality in 4 + 1 strongly coupled plasmas .
these results are in agreement with reference @xcite , where the photon emission rate was calculated starting from the dbi action instead of considering an ads rn black hole .
holographically , turning on the chemical potential means the phase transition from the ads schwarzschild to the ads rn background . in the figures [ fig : figthree ] and [ fig : figfour ]
the ac and dc conductivities are plotted . in the hard wall case ,
the real part of the ac conductivity ( left panel in both figures ) grows monotonically with @xmath102 up to a maximum , then it starts to oscillate . in the case of the soft wall model
, the dilaton prevents that the real part of the ac conductivity begins to oscillate , damping the profile .
in the case of the dc conductivity ( right panels ) , for both models it is increasing with the baryonic charge for a fixed temperature .
this is expected since in the limit @xmath107 we are in the ultrasoft limit , where the chemical potential enhances the photon production by the reasons exposed above .
in this work we have discussed the photon emission rate and the conductivity for hard wall and soft wall model under finite chemical potential regime .
we have constructed holographically the chemical potential by means of an ads reissner nordstrom black hole with a static electrical charge . in the perspective of qft ,
the holographic procedure showed here is equivalent to consider all the feynman diagrams allowed by the landau pomeranchuk
midgal effect that have photons as external legs .
it is also interesting to comment that the results showed here are in agreement with other top / down models used to study photon emission @xcite despite the fact that they are constructed in different backgrounds , as in the case of the sakai sugimoto discussed in @xcite .
the evidence that different holographic approaches can model the same phenomenology encodes an underlying nature : gravity can be connected to strong interactions in a deep way that we are just beginning to uncover .
these models are a good approximation to the study of the excited colored matter created in heavy ion collisions
. an interesting next step would be the description of these ads rn solutions for low temperature , in order to study possible applications to nuclear physics .
99 j. maldacena .
_ large n field limit of superconformal field theories and supergravity_. advacnes in theoretical and mathematical physics , 2 ( 231 ) , 1998 .
d. mateos and l. patio .
_ bright branes for strongly coupled plasmas_. jhep , 2007(11):025 , 2007 . y. y. bu and j. m. yang .
_ structure functions of holographic quark gluon plasma : sakai - sugimoto model versus its non critical version_. phys .
d , 84:106004 , 2011 .
u. gursoy and e. kiritsis .
_ exploring improved holographic theories for qcd : part i. _ jhep , 2008(02):032 , 2008 .
d. l. yang and b. muller .
_ shear viscosities of photons in strongly coupled plasmas_. phys .
b , 760:565570 , 2016 .
a. nata atmaja and k. schalm . _
photon and dilepton production in soft wall ads / qcd_. jhep , 2010(8):124 , 2010 . y. y. bu .
_ photoproduction and conductivity in dense holographic qcd_. phys .
d , 86:026003 , 2012 .
r. a. janik . _
the dynamics of quark gluon plasma and ads / cft _ , p. 147181 .
springer berlin , 2011 .
e. shuryak .
_ the qcd vacuum , hadrons and superdense matter _ , volume 71 , world scientific lecture notes in physics .
world scientific , 2014 .
s. somorendro singh .
_ free energy and direct photon emission at finite chemical potential_. journal of physics : conference series , 535(1):012002 , 2014 . h. gervais and s. jeon .
_ photon production from a guark gluon plasma at baryon chemical potential_. phys .
c , 86:034904 , 2012 .
p. arnold , g. d. moore and l. g. yaffe .
_ photon emission from ultrarelativistic plasmas_. jhep , 2001(11):057 , 2001 .
p. arnold , g. d. moore and l. g. yaffe .
_ photon emission from quark gluon plasma : complete leading order results_. jhep , 2001 ( 12):009 , 2001
. s. c. simon , p. kovtun , g. d. moore .
_ photon and dilepton production in sypersymmetric yang
mills plasma_. jhep , 2006(12):015 , 2006 . j. polchinski and m. j. strassler .
_ hard scattering and gauge / string duality _ phys .
, 88:031601 , 2002 . h. boschi filho and n. r. f. braga . _ qcd / string holographic mapping and glueball mass spectrum_. eur .
j. c , 32(4):529533 , 2004 .
a. karch , e. katz , d. t. son and m. a. stephanov .
_ linear confinement in ads / qcd_. phys .
d , 74:015005 , 2006 . k. jo and s. j. sin .
_ photoemission rate of strongly interacting quark gluon plasma at finite density_. phys .
d , 83:026004 , 2011 .
s. w. hawking and h. s. reall .
_ charged and rotating ads black holes and their cft duals_. phys .
d , 61:024014 , 1999 .
r. g. cai and j. p. shock .
_ holographic confinement / deconfinement phase transition of ads / qcd in curved spaces_. jhep , 2007(08):095 , 2007 .
p. colangelo , f. giannuzzi and f. zuo .
_ temperature and chemical potential dependence of the gluon condensate : a holographic study_. phys .
d , 88:115011 , 2013 . l. patio and d. trancanelli .
thermal photon production in a strongly coupled anisotropic plasma .
jhep , 2013(2):154 , 2013 .
d. t. son and a. o. starinets .
_ minkowski space correlators in ads / cft correspondence : recipe and applications_. 2002(09):154 , 2013 . s. nakamura .
_ comments on chemical potentials ads / cft_. progress of theoretical physics , 119(5):839849 , 2008 . y. seo and s. j. sin .
_ baryon mass in medium with holographic qcd_. jhep , 2008(04):010 , 2008 . s. nakamura , y. seo and k. p. yogendran , _ baryon - charge chemical potential in ads / cft_. progress of theoretical physics , 120(1):5176 , 2008 .
c. p. herzog , _ holographic prediction for the deconfinement temperature_. phys .
, 98:091601 , 2007 . n. r. f. braga , s. diles and m. a. martin contreras .
_ holographic model for heavy vector
meson masses_. epl ( europhysics letters ) , 115(3):31002 , 2016 . i. kanitscheider , k. skenderis and m. taylor .
_ precision holography for non
conformal branes_. jhep , 2008(09):094 , 2008 . s. endres , h. van hees and m. bleicher .
_ photon and dilepton production at the facility for antiproton and ion research and beam - energy scan at the relativistic heavy ion collider using coarse grained microscopic transport simulations_. phys . rev .
c , 93:054901 , 2016 .
d. dutta , k. kumar , a. k. mohanty and r. k. choudhury .
_ chemical equilibration and thermal dilepton production from the quark gluon plasma at finite baryon density_. phys .
c , 60:014905 , 1999 .
p. aurenche , f. gelis , h. zaraket and r. kobes .
_ bremsstrahlung and photon production in thermal qcd_. phys .
d , 58:085003 , 1998 .
p. aurenche , f. gelis , g. d. moore and h. zaraket .
_ landau pomeranchuk midgal resummation for dilepton production_. jhep , 2002(12):006 , 2002 . | we calculate the photon emission rate and the electrical conductivity of the qgp at finite temperature and finite chemical potential using ads / qcd approximations in an ads reissner nordstrom background .
to do so , we supposed the medium properties to be encoded in a geometric background .
the results obtained in the hard wall and soft wall model are consistent with the observed phenomenology and they also in agree with other holographic results , as the d3/d7 or the sakai sugimoto models , suggesting the universality of ads / cft conjecture as tool to explore qcd . |
the spectroscopic detection of an energetic supernova ( sn2003dh ) concurrent with grb 030329 ( hjorth 2003a ; stanek 2003 ) has firmly established that long - duration ( @xmath2 s ; kouveliotou 1993 ) grbs accompany the core - collapse of some class of massive stars . since
massive stars are short - lived , this also confirms that grbs are closely associated with star formation activity ; a possibility already discussed by a number of authors ( eg . wijers 1998 ; totani 1999 ; blain and natarajan 2000 ) .
the extreme luminosity of the prompt gamma - ray emission means that grbs can be detected , if they exist , to very high redshifts , with minimal extinction by intervening gas or dust .
this makes them potentially very powerful indicators of star formation to early times . to date
, spectroscopically - confirmed redshifts have only been published for about three dozen bursts , although various schemes have been suggested to derive redshifts empirically from @xmath3-ray properties ; such as the lag - luminosity ( norris 2000 ) and variability - luminosity relations ( reichart 2001 ) .
these studies suggest that the redshift distribution of grbs is broadly consistent with the emerging picture of the comoving star formation rate in the universe having peaked sometime around redshifts 1 - 4 , although uncertainties ramp up at higher redshifts .
( eg . ramirez - ruiz , trentham and blain 2002 ; lloyd - ronning 2002 ) these results are interesting , but premature to the extent that we have limited knowledge of these luminosity correlations , and only rudimentary understanding of the relationship between grb rate and star formation rate ( eg .
krumholz 1998 ) .
for instance , there is the possibility that grb rate and/or brightness depends also on other factors , such as metallicity , galactic environment , and certainly on any variations in the stellar initial mass function ( imf ) .
these are difficult factors to disentangle , but important insights can be gained by comparing grb rate with the star formation rate estimated by other means .
our program is aimed at providing a more quantitative comparison of the star formation rate deduced by grbs and that obtained from submm luminosity , through direct study of the host galaxies of grbs .
existing methods of mapping the star formation history of the universe rely on estimating the star formation rates of individual galaxies , and summing these up in redshift bins with some estimated correction for galaxies below the detection threshold ( eg .
adelberger & steidel 2000 ) , or by modelling the redshift distribution so as to fit integrated backgrounds ( eg .
blain 1999a ) .
if grbs reliably trace star formation , and we can quantify the relationship , then ultimately detection of their host galaxies will not be required for the purposes of mapping global star formation history .
nonetheless , grb hosts also uniquely allow us to study the star - forming galaxy luminosity function right to the faint end .
an upshot may be a means to estimate the proportion of the total star formation which is going on in ir - bright , dusty galaxies , optically - bright galaxies and in fainter populations which are not selected in other surveys ( eg .
trentham 2002 ) . to date , the best example of the power of grb selection to probe the faintest end of the galaxy luminosity function is grb 020124 , whose host galaxy is undetected by _
hubble space telescope ( hst ) _ to @xmath4 ( berger 2002 ) .
this galaxy would not have been found in any direct imaging survey to date and yet observations of the afterglow show that the host has a high hi column density ( making it a damped ly@xmath5 absorber ) and a redshift of @xmath6 ( hjorth 2003b ) .
the power of submm studies for mapping the star formation history of the universe is discussed in detail by blain ( 2002 ) and smail ( 2002 ) .
briefly , in dusty systems uv radiation , predominantly from massive stars , is reprocessed by the dust and emitted in the far ir .
this emission , which itself is unaffected by dust extinction , is thus proportional to the obscured star formation rate . at higher redshifts
the peak of emission moves increasingly into the submm , with the beneficial consequence that at 850 the apparent luminosity of a galaxy of given intrinsic bolometric luminosity changes little from redshift @xmath7 out to @xmath8 .
submm surveys , combined with constraints from the intensity and spectrum of ir backgrounds , show that compared to the local universe , the majority of high redshift star formation appears to be taking place in dusty systems ( blain 1999a ) , and much ( although not the dominant part ) of this in so - called ultraluminous infrared galaxies ( ulirgs : with ir luminosities @xmath9 ) .
low redshift ulirgs exist , but have around 1000 times lower comoving space density ( smail 2002 ) .
potential drawbacks with submm surveys are : ( a ) at the 2 mjy confusion limit of scuba only about 30% of the total submm background emission from _
cobe_-firas observations is resolved out ; ( b ) any individual galaxy may suffer some contamination of its 850 luminosity through heating from an obscured agn . only about 10% of submm galaxies with deep x - ray data appear to show evidence for a hard x - ray agn (
almaini 2003 ) , but larger samples and deeper observations will be necessary to confirm this fraction ; ( c ) it is necessary to assume or constrain the shape of the far - ir spectral energy distribution ( sed ) at wavelengths shorter than 850 in order to translate the measured submm flux density into an accurate luminosity and star formation rate .
the luminosity inferred from a galaxy with a certain sed depends only weakly on redshift ; however , without direct knowledge of the details of the sed this luminosity is uncertain , leading to ambiguity in the results .
a certain fractional change in the dust temperature leads to a fractional uncertainty in the inferred luminosity that is greater by several times ( blain 2002 ; blain , barnard and chapman 2003 ) .
while results are so far generally consistent with dust temperatures of order 35k ( chapman 2003a ) , the extent of the distribution of values is not yet known .
redshifts for large samples of submm galaxies were for a long time hard to obtain , because of the lack of both bright optical counterparts ( frequently @xmath10 ) , and the poor 15 arcsec spatial resolution of scuba at 850 . however , chapman et al .
( 2003b ; see also ivison 2002 ) have demonstrated that about 65% of submm galaxies
brighter than 5mjy are detected in very deep vla radio maps , providing accurate subarcsec positions .
redshifts for around 4050% of these radio - detected submm galaxies can be obtained using keck lris spectra ( chapman 2003a ) .
based on these results they find a range of redshifts from 0.8 to 2.8 with a median of 2.4 , and conclude it is likely that most submm galaxies lie at redshifts between 2 and 3 . the radio selection could be biased to lower redshifts , and to sources containing agn ; however , the reasonable @xmath11% completeness of chapman s redshift determinations suggests that these effects are not too significant . on the other hand , in a recent study of the evolution of the global stellar mass density @xmath12 based on an infrared - selected sample of galaxies from the hubble deep field north , dickinson et al .
( 2003 ) report that while the star formation rate essentially tracks that determined in other wave - bands at low redshifts , the rate at @xmath13 is significantly different .
these observations appear to be inconsistent with scenarios in which the bulk of stars in present - day galactic spheroids formed at @xmath14 , since most of the stars ( 50 75% ) of the present day stellar mass density formed by @xmath15 and in fact by @xmath16 only 3 14% of today s stars were present .
the issue is clearly not settled .
current sensitivities and confusion limits imply that substantially fainter submm samples will not be studied in detail until the alma interferometer is commissioned in 2012 .
hence , an alternative method to probe the star formation in obscured galaxies is definitely required , and grb source counts and host galaxy studies may ultimately be able to tell us both when and where this star formation occured . several characteristics of grbs lend them to probing high redshift star formation : ( a ) they are bright enough to be seen to @xmath17 ( eg .
lamb & reichart 2000 ) ; ( b ) they can be detected in @xmath3-rays through large columns of dust and gas ; ( c ) they can be detected independently of whether a host galaxy can be found .
furthermore , much information relating to the host , such as redshift , metallicity , gas column density and extinction can all be obtained indirectly , from afterglow observations
either optically or x - ray ; ( d ) the spectral slopes of both prompt and afterglow emission compensate to some extent for redshift dimming , and time - dilation means that observations can be made earlier in the rest - frame time than would be the case for lower redshifts ; ( e ) being produced by individual massive stars , they are obviously unaffected by agn contamination . on the downside , grbs are rare , their progenitors are still not fully understood and they are not very useful for telling us about the star formation rate in individual galaxies . in terms of the current observational state - of - play , samples of grbs , particularly those with firm redshifts are very inhomogeneous the result of a wide variety of triggers and followup campaigns .
one consequence is that , of the roughly 200 entries to - date in jochen greiner s web tablejcg / grbgen.html ] of `` well - localised '' bursts , fewer than 25% have optical afterglows identified .
however , only a minority of those 200 were reported within 24 hours , and had error circles @xmath18 arcmin in diameter . also ,
many were not well placed for optical observation , have only shallow limits , or in some cases no reported optical followup at all .
thus , the superficially high rate of `` dark '' grbs is misleading .
the proportion of genuinely dark grbs ( a reasonable working definition would be @xmath19 at 24 hours post - burst ; see also section 2 ) amongst the _ hete _ and _ bepposax _ triggers is probably only 1030% .
the wide variation in grb followup campaigns makes it hard to quantify selection effects , but selection effects may well be important for our study . to be found optically
, an afterglow should not be in too dusty an environment .
a low redshift illustration of this issue is provided by mannucci ( 2003 ) , who find an enhanced core - collapse supernova rate in k - band monitoring of nearby star - forming galaxies , but still conclude that the large majority of supernovae remain undetected due to very high extinction .
in fact , for grbs this conclusion is not as inevitable as it sounds because the initial flux of high energy photons from grbs is expected to destroy dust possibly up to @xmath20 pc ( eg . fruchter 2001c ; galama and wijers 2001 ) . in some cases this will be enough to create a window through otherwise obscuring dust clouds .
of course , if most star formation in the universe is occuring in the `` obscured mode '' , then dust destruction may actually be a _ requirement _ to explain the large number of afterglows detected optically . even when optical afterglows are not found to faint limits , as we shall argue below , it is not necessarily the case that they are heavily enshrouded in dust . nonetheless , it would be surprising if some fraction of grb afterglows were not missed because they were highly extinguished .
detected afterglows are unlikely to be found in very low - density environments , even if grbs occur there , since the brightness of the spectral peak reduces with the density of the ambient medium ( sari 1998 ) .
the high hi column densities seen towards many bursts ( eg .
galama & wijers 2001 ; hjorth 2003b ) are consistent with this expectation , although reichart & price ( 2002 ) suggest that for the limited sample of dark bursts available , the column densities are not consistent with them being in the nuclear regions of ulirgs either .
observational limitations are also likely to introduce a bias against finding high redshift grbs .
although the spectral slope and cosmological time - dilation work so as to reduce the effect of redshift on the magnitude of an afterglow at a given observer time after the burst ( another kind of negative @xmath21-correction ) , grbs of comparable intrinsic luminosity will still appear fainter at higher redshift , making them and their afterglows harder to discover . for example , hogg and fruchter ( 1999 ) adopted a @xmath22 dependence for the probability that a burst at a given redshift is detected and an optical afterglow found .
of course , @xmath23 objects will be essentially invisible in the optical due to the ly@xmath5 break .
our initial goal is to compare the submm properties of grb hosts with model predictions and hence provide a consistency check on both techniques for tracing star formation .
ultimately we d like to understand the quantitative relation between grb rate and star formation rate , so that larger samples of grbs may be used to give a good description of the global star formation history of the universe .
the first submm limits and detections of grb host galaxies came from observations aimed at detecting grb afterglows . only in a couple of cases were such fading afterglows detected , but grb 010222 was found to be a steady source , suggesting the flux was dominated by emission from a bright host galaxy ( frail 2002 ) .
useful upper limits were also obtained for a number of other hosts ( smith 1999 ; galama 1999 ; kulkarni 1999b ; smith 2001 ) .
it is possible to average a number of non - detections providing systematic uncertainties are small , to find the average flux density for a whole sample .
unfortunately in the case of the afterglow observations , since there may well be some small afterglow contribution to each observation , even if it is not detected significantly in individual cases , this averaging procedure is best avoided . for the data discussed here the observations
were generally made long after the afterglow should have faded .
an obvious concern , already raised in section 1.3 , is that those grbs with optically detected afterglows may be biased against residing in dusty hosts , the very galaxies which would on average be submm bright . in an effort to assess whether such a bias exists ,
barnard ( 2003 ; hereafter paper 1 ) observed a small sample of `` dark '' grbs with deep limits on any optical afterglow but with good radio and/or x - ray positions .
the expection at the outset was that compared to the hosts of optically - bright grbs , these would be more likely to be dusty , massive star - forming galaxies .
in fact , this sample of 4 produced only 1 individually significant detection , and no overall excess of 850 emission over that predicted by ramirez - ruiz , trentham & blain ( 2002 ) for all hosts .
these results appear to argue against dark bursts being preferentially found in dusty hosts .
however , this is only a small sample , and in paper 1 we also remarked that there was some evidence from the rapid decay of the radio afterglows that two of the bursts in submm - faint hosts could have been dark due to intrinsic optical faintness , rather than extinction .
in fact , it is becoming clearer that there is a broad range in brightness of optical afterglows and a number of detected bursts would have been missed in most afterglow searches . for example , grb 980613 , grb 000630 and grb 021211 ( hjorth 2002 ; fynbo 2001 , fox 2003 ) were all around @xmath24 at 1 day , and , although the sample of bursts studied in paper 1 were fainter than all these , the upper limits do not require them to have been much fainter .
similarly , grb 020124 had a relatively typical intrinsic magnitude , but appeared faint because of its redshift of @xmath25 ( hjorth 2003b ) . in any event
, as we see below , even the modest rate of submm detections for dark burst hosts found in paper 1 appears to be somewhat greater than the rate which is now found for hosts of optically detected bursts .
recently berger ( 2003 ; hereafter be03 ) published scuba 850 measures for a larger sample of 13 grb hosts , in addition to radio observations for many .
below we present our results for another sample of hosts ( which partially overlaps with be03 ) , and combine and analyse all the extant data .
we have obtained further submm photometry of the host galaxies of optically identified grbs , using the sub - millimetre common user bollometer array ( scuba ) instrument on the james clerk maxwell telescope ( jcmt ) .
the targets were chosen to be well placed for observation from mauna kea , and have sub - arcsec positions from their optical afterglows .
all observations were made at least 12 weeks after the burst ( and usually much longer ) , and are therefore very unlikely to be contaminated with afterglow emission .
observations and reduction were performed as described in paper 1 .
the log and results are detailed in table 1 ; note that some were reported by barnard ( 2002 ) .
of the galaxies presented here and in paper 1 , seven are in common with be03 .
this is useful to improve the measurement uncertainties , and also to check for statistical consistency , since the methods of reduction differ somewhat .
overall , we are reassured to find no significant systematic difference between the two data sets . the galaxy to galaxy scatter is , however , a little larger than expected from the quoted errors .
specifically , 4 of the 7 disagree by more than @xmath26 , which could happen by chance , but it leads us to suspect that the uncertainties found by one or both groups are marginally underestimated . .
850photometry observations of a new sample of grb hosts , chosen on the basis of good positions easily observable from mauna kea .
note the varying number of integrations per target the quoted flux densities are error - weighted averages of the measurements for each source .
the @xmath27 measures the sky opacity during the observations .
as can be seen , we do not enforce positivity on the submm fluxes , so as not to bias any subsequent statistical analysis . in a few cases the uncertainties are still large , and the data of limited use , but we report them here for completeness . [ cols="<,^,>,^,>",options="header " , ]
the combined sample contains 21 host galaxies which have 850 measures with better than 1.4 mjy uncertainties , indeed most of these have errors @xmath28 1 mjy .
our discussion refers mainly to this well - observed sample .
only three galaxies have significant positive detections , namely the hosts of grb 010222 , grb 000418 and grb 000210 , which are all found at @xmath29 , and so can be regarded as confident detections .
a few others have @xmath30 detections , but we should be more wary of these , particularly since there are also a couple of cases of @xmath30 negative fluxes .
this latter fact could be taken as a further hint of a small underestimate in the errors , but equally it could be a chance occurence given the sample size and the fact that the quoted errors do not account for crowding noise .
for reasonable dust temperatures of order 40k , corresponding to restframe far - ir seds peaking at about 90@xmath31 m , the luminosities of these galaxies would be about @xmath32l@xmath33 .
if all this energy was provided by star formation , then a star formation rate upward of 1000m@xmath33yr@xmath34 would be required ( eg .
blain , barnard and chapman 2003 ) .
two hosts previously identified as possibly highly star forming are grb 980703 and grb 000911 , on the basis of radio flux ( berger 2001b ) and submm ( be03 ) respectively .
our compilation suggests that in both instances , these initial results were overestimates in the former case , plausibly due to the scatter in the fir - radio correlation . in the case of grb 000911 , the conclusion proceeds from the low flux seen in the afterglow observations of smith ( 2001 ) . as a general point
, we note that when working at the limit of detection , since there are many more low luminosity galaxies than high luminosity , the fluxes of the brightest galaxies in any sample are more likely to be overestimates than underestimates . this effect is akin to to eddington - malmquist bias , and tells us that we should not be surprised if the fluxes of the brighter galaxies are frequently found to be lower on remeasurement . for these 21 hosts , the error - weighted mean 850 flux density is @xmath35 mjy .
this can be used to test the null hypothesis of zero flux for the sample , and , as expected , rejects it with confidence . however , this number is not a fair estimate of the true mean of the sample since we have not accounted for the intrinsic dispersion in grb host galaxy luminosities ( about which we do not have prior knowledge ) , and also because the brighter galaxies were in some cases ( notably grb 010222 ) observed for longer and hence have smaller error bars for that reason .
the unweighted mean is @xmath36 mjy , and is a fairer estimate of the true mean .
this is higher than the average 850 flux density found for samples of lyman - break galaxies , which range between 0 and @xmath370.4 mjy , depending on the exact sample selection ( chapman 2000 ; peacock 2000 ; webb 2003 ) .
what distribution do we expect to see ? as a starting point , we assume that both submm luminosity and grb rate are perfectly correlated with star formation rate . in that case , if submm flux and grb detection completeness were also perfectly redshift independent , then the predicted distribution would simply be the submm - luminosity - weighted luminosity function of all galaxies .
ramirez - ruiz , trentham & blain ( 2002 ) use the models of blain ( 1999a , b ) for the evolution of the submm galaxy population to predict the the 850 flux density distribution of grb hosts .
uncertainties in the model are almost entirely due to the uncertain link between the grb rate and high - mass star formation heating dust .
subject to the , probably low , rate of contamination in the submm galaxy population from agn heating , and to possible evolution of the initial mass function ( imf ) , the flux density distribution of submm galaxies is reasonably well constrained throughout the interval 115mjy ( blain 2002 ; borys 2003 ) . although it predates the substantially complete observed redshift distribution of submm galaxies from chapman ( 2003a ) , the assumed model is in good agreement with these results .
it is possible that the galaxies are systematically cooler and less luminous ( eg .
efstathiou & rowan - robinson 2003 ) , however , if that is the case , then the redshift distribution is only consistent with significant evolution of the form of the far - infrared radio correlation with redshift ; figure 1 shows the predictions of this model as filled points . to provide a fair comparison ,
the model output was convolved with with the distribution of observational errors , and includes a correction to allow for the possibility of the chance appearance of an unrelated submm source in the scuba beam .
the latter correction amounts to roughly 0.7 of an object contaminating the 24 mjy bin and is based on the source counts summarised in blain ( 2002 ) .
we have chosen to put error bars on the model , reflecting the counting statistics of the proportion expected in each bin .
the histogram represents the submm observations .
it is apparent that there is a dearth of hosts with @xmath1 mjy 850 flux densities , which is becoming statistically significant ( formally , at the 1.6@xmath38 level ) .
in fact , the significance is greater if we consider the remaining 5 galaxies with larger error bars , namely grb 970828 , grb 990308 , grb 001025a , grb 020124 and grb 020813 ( and indeed one might also include grb 980519 from smith 1999 ) , since at @xmath26 these are all inconsistent with being @xmath1 mjy sources as well . furthermore , we note that if the estimated flux uncertainties are somewhat optimistic ( as suggested in section 3 ) , it tends to make the discrepancy between theory and observations rather worse .
this seems counterintuitive , but is due to the fact that larger observational errors should result in more low luminosity systems appearing in the brighter bins by chance . at this stage
there are many possible explanations for the discrepancy ( beyond the relatively small sample size ) .
first and foremost , we must still worry that amongst the grbs with `` very dark '' afterglows , for which good radio or x - ray positions are not available , lurk a small number of highly submm luminous hosts .
another potential selection effect is that grbs are being preferentially picked up at lower redshifts , both in terms of initial detection , but more significantly the detection of the afterglows , as discussed in section 1.3 .
this would be a less direct selection against the higher redshift , dusty systems , although we note that of the seven systems with @xmath39 , _ none _ have significant 850 fluxes .
other possible physical explanations are : ( a ) that grbs are preferentially found in low metallicity smaller and less dusty systems .
this has been predicted as a consequence of the single - star collapsar model ( heger 2003 ) , and is consistent with the apparent enhanced brightness of grbs in the outer parts of their parent galaxies ( ramirez - ruiz , lazzati & blain 2002 ) , and the relatively low metallicity inferred for many grb hosts ( eg . fynbo 2003 ) .
the latter study notes that a a large proportion of grb hosts are ly@xmath5 emitters ( five out of five so far studied ) , which is significantly greater than the proportion of lyman - break galaxies , although perhaps surprisingly , many of the scuba galaxies for which chapman have recently acquired redshifts , are also strong ly@xmath5 emitters ; ( b ) that the higher luminosity submm galaxies are more contaminated by agn , or the submm luminosity function more contaminated by cooler , less luminous galaxies , than has been generally thought ; ( c ) that whilst both grbs and submm flux trace star - formation rate , the two are not perfectly correlated due to different phase lags with respect to the true star formation rate .
for instance , the simulation of bekki and shioya ( 2001 ) of a merger induced starburst , shows significant fluctuations in the star - formation rate on timescales much less than 100 myr ( their fig 6 ) .
given that both grb rate and submm flux will not follow rapid fluctuations instantaneously , some decorrelation may occur .
( d ) that variations in the stellar imf are having a different effect on grb rate compared to submm flux ; or ( e ) that grbs for some reason are occuring preferentially in galaxies with high dust temperatures , so more of the bolometric luminosity appears at shorter wavelengths and is being missed at 850 .
if submm flux traces star formation , and the selection of grb hosts is unbiased , then the submm - bright grb hosts as a whole should be similar to the general population of submm bright galaxies .
of course , now we are dealing with a very small sample , but it is interesting to examine the properties of the three hosts with 850 flux densities above 2 mjy .
gorosabel ( 2003a ) have constructed a uv to ir sed for the host to grb 000210 , which is well fit by a starburst sed with a relatively modest star formation rate of 2 and negligible extinction . the redshift is @xmath40 and the luminosity @xmath41 .
the uv to ir sed of the grb 000418 host ( gorosabel 2003b ) is also well fit by a starburst template , in this case with a moderate amount of extinction @xmath42 , and a star formation rate up to 60 .
a similar result is obtained in the spectroscopic study of bloom ( 2003 ) , who also find no evidence of any agn contamination .
this star formation rate is still an order of magnitude less than that derived from the submm , but that is not unreasonable for heavily obscured galaxies .
the redshift is @xmath43 and the luminosity is @xmath44 .
it is compact with effective radius @xmath45 kpc , and the afterglow was located close to the optical centroid ( fruchter & metzger 2001 ; bloom 2003 ) .
the afterglow itself has produced extinction estimates ranging from @xmath42 ( berger 2001a ) to @xmath46 ( klose 2000 ) .
the host of grb 010222 is fainter still in the optical at @xmath47 ( fruchter 2001a ) , but it is also moderately blue , with @xmath48 and an intrinsic luminosity @xmath49 ( frail 2002 ) .
the redshift is @xmath50 and again _ hst _ images reveal the afterglow to have been located on top of the peak of the optical emission . in all these cases
if the submm is really indicating copious amounts of hidden star formation activity , it would appear that the optically visible part of the host is dominated by one or more relatively unextinguished regions .
this is plausible it could be concentrated in a shell or plume of material , or possibly a number of `` windows '' through an obscuring shroud ( eg .
bekki 1999 ) . however , if this is the case , it also seems that the optically bright regions must be spatially proximate to the bright far - ir emission if we are to understand why the afterglows ( as with most detected grbs ) tend to be found close to the optical centroids ( or hotspots ) of the hosts .
so , how do the properties of these galaxies compare with the the optical / ir counterparts of the submm bright galaxies selected in blank - field scuba surveys ?
in fact the bright scuba galaxies display quite a wide variety of characteristics .
they have been split into three classes ( smail 2002 ) : class 0 are relatively bright in optical as well as submm much of the star formation is unextinguished ; class 1 are extremely red objects ( eros ) with @xmath51 typically greater than 5 ; finally class 2 are extremely faint in optical and ir .
this classification system was conceived based on galaxies which are generally brighter than 5 mjy at 850 .
although no grb host to - date is that bright , it appears the the submm brightest of the grb hosts do not fit neatly into this scheme , being intermediate in optical / ir luminosity between classes 0 and 2 , but not very red eros like class 1 .
interestingly , the best candidate for an extremely red grb host galaxy , that of grb 980326 which has @xmath52 ( although the @xmath53 detection at @xmath54 is @xmath55 ; chary , becklin & armus 2002 ) , has a low submm luminosity .
this compares to an average 850 flux density of @xmath56 mjy recently found by wehner , barger & kneib ( 2002 ) for a @xmath53-selected sample of eros .
furthermore , as discussed in section 1.2 , the redshift distribution of bright scuba sources shows a peak in the redshift range 23 , whereas these 3 submm bright grb hosts are nearer @xmath15 .
we conclude that the properties of the few submm bright grb hosts found to - date are not very representative of the submm - bright galaxies as a whole : they do not fit easily within any of the 3 classes .
be03 reached the same conclusion based on a colour redshift plot , in which the submm - bright grb hosts lie bluer and at lower redshift than the submm selected galaxies as a whole .
this could be because the grb hosts are somewhat fainter at 850 , since nearly all well studied submm samples are @xmath57 mjy ( although frayer 2003 report a detailed study of a single gravitationally lensed source at @xmath58 , which has a low submm flux density of @xmath59 mjy after correction for lensing ) .
as discussed above , we also expect grb selection to provide some bias toward picking up lower redshift galaxies , and against very dusty galaxies .
the alternative , of course , is that either the submm selection has some problems , such as a surprising rate of agn contamination , or surprising dust properties , or grb progenitors are more likely to arise in less dusty systems .
we have shown that whilst a small number of grb hosts are bright submm galaxies , indicating high star formation rates , the proportion , particularly at flux densities @xmath1 mjy , is fewer than predicted if both grbs and submm flux trace star formation in an unbiased way . furthermore , and similarly puzzling , the three submm - brightest grb hosts are not very typical in terms of their optical properties and redshift distribution of the galaxies selected in blind submm surveys .
of course , the current sample size is still rather small , and very likely suffers some selection biases . in particular , grbs in very dusty hosts and/or at higher redshifts
are more likely to be missing from the observed afterglow samples .
a small number of these amongst the `` dark '' bursts ( those without optical counterparts to deep limits ) could remove the discrepancy although the afterglows known to be genuinely very faint is only a small fraction of those for which optical emission has not been detected .
furthermore , a fair proportion of the darker bursts seem to be dark for reasons other than their residing in high redshift , dusty ulirgs , so it s not obvious that this selection effect provides the full explanation .
future studies may largely overcome these selection problems by much earlier and more uniform afterglow searches with _ swift _ and various robotic telescopes
. the positions from the _ swift_/xrt may be good enough to identify ulirg hosts , even when no optical afterglow is detected .
_ spitzer _ will also be a powerful tool , particularly the for the lower redshift hosts @xmath60 , where the large majority of the bolometric luminosity appears in the far - ir .
if it turns out that the discrepancy with the model predictions persists when more complete samples are studied , it will certainly be telling us something interesting about the astrophysics of grbs and/or the relationship between fir emission and star formation . on the other hand ,
if the disagreement goes away , the use of grbs as practical star formation indicators will be strongly bolstered .
we thank the referee , steve serjeant , for insightful comments , jochen greiner for keeping his invaluable website up to date ( http://www.mpe.mpg.de/@xmath37jcg/grbgen.html ) and paul vreeswijk , jens hjorth , javier gorosabel and johan fynbo for useful discussions .
we also thank all the observers and staff at the jcmt who carried out flexible - scheduled observations on our behalf .
the jcmt is operated by the joint astronomy centre on behalf of the particle physics and astronomy research council of the uk , the netherlands organization for scientific research , and the national research council of canada . | long duration gamma - ray bursts ( grbs ) accompany the deaths of some massive stars and hence , since massive stars are short lived , are a tracer of star formation activity . given that grbs are bright enough to be seen to very high redshifts , and detected even in dusty environments , they should therefore provide a powerful probe of the global star formation history of the universe .
the potential of this approach can be investigated via submm photometry of grb host galaxies .
submm luminosity also correlates with star formation rate , so the distribution of host galaxy submm fluxes should allow us to test the two methods for consistency . here , we report new jcmt / scuba 850 measurements for 15 grb hosts . combining these data with results from previous studies
we construct a sample of 21 hosts with @xmath0 mjy errors .
we show that the distribution of apparent 850 flux densities of this sample is reasonably consistent with model predictions , but there is tentative evidence of a dearth of submm bright ( @xmath1 mjy ) galaxies .
furthermore , the optical / infrared properties of the submm brightest grb hosts are not typical of the galaxy population selected in submm surveys , although the sample size is still small .
possible selection effects and physical mechanisms which may explain these discrepancies are discussed .
stars : evolution dust , extinction galaxies : evolution cosmology : observations gamma - rays : bursts infrared : galaxies . |
the generation of quantum effects in nano - mechanical cavities with movable mirrors has been the subject of a great interest in recent years @xcite .
this interest stems from the possibility of the development of new practical techniques for cooling macroscopic objects to very low temperatures and for engineering of entangled states of macroscopic systems . with the recent progress in laser cooling techniques , fabrication of low - loss optical elements and high-@xmath0 mechanical resonators
, it is now possible to prepare nanomechanical oscillators that can be controlled to a very hight precision and can even reach the quantum level of the oscillations @xcite . in these systems , the vibrations of mechanical oscillators are induced by radiation pressure that creates a strong nonlinear coupling of the vibrational mode to radiation modes .
most of these studies have been done on examples provided by cavity optomechanical systems with linear or ring cavities @xcite .
it has been demonstrated that a large radiation pressure can be generated in the cavity that in return may lead to entanglement between different components of the system . in particular ,
stationary entanglement has been predicted between the cavity mode and a vibrating mirror @xcite , between an atomic ensemble or a bose - einstein condensate located inside an optical cavity and the vibrating mirror of the cavity @xcite , between two vibrating mirrors of a ring cavity @xcite , between two dielectric membranes suspended inside a cavity @xcite , and between a membrane and a trapped atom both located inside a cavity @xcite .
further studies have addressed interesting problems of entangling mechanical oscillators @xcite , entangling optical and microwave cavity modes @xcite , and the creation of a photon by a vibrating mirror @xcite . in this connection
, we should mention the most recent work on the generation of entanglement in pulsed cavity optomechanics @xcite , and the work on the creation of entanglement between two oscillating mirrors through the coupling of the mirrors to an atomic system @xcite .
the purpose of this paper is to explore coherence and correlation features of an optomechanical system composed of three bosonic modes realized with an one - dimensional optical lattice located inside a single - mode cavity with a movable mirror .
we are particularly interested in the effect of the first - order correlation ( first - order interference ) on entanglement between two modes which is associated with second - order correlation functions .
it is well known that in the parametric down - conversion , served as a typical source of entanglement , the signal and the idler beams are strongly entangled but always behave as mutually incoherent @xcite .
a similar conclusion applies to the correlations between modes of the optomechanical system , where it was demonstrated that the cavity and mechanical modes play the same role as the signal and the idler of a nondegenerate parametric oscillator and the modes behave as mutually incoherent @xcite .
this seems to suggest that entanglement between two modes rules out the first - order coherence between them .
we consider various situations where the modes of the optomechanical system can be made mutually coherent and to exhibit the first - order interference .
this leads to an obvious question of to what extent of the first - order coherence could affect entanglement between the modes . to address this question , we use a polariton model of the optical lattice coupled to a single - mode cavity field and calculate various coherence and correlation functions of the optomechanical system .
we show that the system is capable of generating a wide class of coherence and correlation effects , ranging from the first - order coherence , the anomalous autocorrelations and anomalous cross correlations between the modes . in a series of simple examples
we show that the generation of the first - order coherence between two modes of the system is equally effective in destroying entanglement between these modes .
we illustrate our considerations by examining two and three mode interactions . after establishing the connection between the generation of entanglement and the first - order coherence
, we consider the problem of the creation of entanglement between two independent modes by coupling them to an intermediate mode .
we show that the coupling of the modes to the intermediate mode by a parametric interaction results in no entanglement between the modes . we then consider a different coupling configuration and find that an entanglement can be generated between two independent modes if one of the modes is coupled to the intermediate mode by the parametric interaction and the other is coupled by the linear - mixing interaction .
the paper is organized as follows .
we begin in sec .
[ sec2 ] with the description of the model .
we represent the finite size optical lattice that is located inside a single - mode cavity in terms of bloch - type waves called excitons , and diagonalize the interaction hamiltonian of the excitons plus the cavity mode to describe the system in terms of bright polaritons .
we then proceed in sec .
[ sec3 ] to study the dynamics of the system in terms of the quantum langevin equations for relevant variables .
we use the linearization approach to the equations of motion and arrive to a set of three couple differential equations for the fluctuation operators , which we solve for the steady - state .
we apply the solution in sec .
[ sec4 ] to the calculation of the bipartite coherence and correlation functions of the polariton and the mechanical modes .
we discuss the conditions for entanglement in terms of the squeezing fluctuations and the cauchy - schwartz inequality .
the anomalous autocorrelation and cross correlation functions are introduced to discuss conditions for the violation of the cauchy - schwartz inequality .
a possibility of generating a two - color entanglement is also discussed . in sec .
[ sec5 ] we evaluate the first - order coherence and entanglement in the case of three coupled modes .
two coupling configurations of two independent modes to the intermediate mode are discussed . in sec .
[ sec6 ] , we examine parameter ranges in which the predicted coherence and correlation effects could be observed with the current experiments . finally , in sec .
[ sec7 ] we summarize our results .
[ sec2 ] we consider a finite size one - dimensional optical lattice located inside a single - mode cavity with one fixed partially transmitting mirror and one movable perfectly reflecting mirror , as shown in fig .
[ mirfig1 ] .
the optical lattice is composed of @xmath1 regularly spaced and non - overlapping sites located at positions @xmath2 , with total length @xmath3 , where @xmath4 is the separation between the sites and @xmath5 is the cavity mode waist at the position of the lattice @xcite .
the lattice is formed by two counterpropagating laser beams entering the cavity from the sides and forming a standing wave in the direction perpendicular to the direction of the cavity mode .
the cold atoms loaded on the optical lattice are confined in an array of microscopic trapping potentials , forming a mott - insulator - like medium with one atom per site @xcite .
the atoms in the optical lattice are modeled as two - level systems with ground state @xmath6 and excited state @xmath7 , separated by the transition frequency @xmath8 and connected by a transition dipole moment @xmath9 , which can be assumed to be real with no loss of generality . since the optical lattice is formed in the direction perpendicular to the cavity axis , the effect of the motion of the atoms ( center of mass motion ) on the coupling strength of the atoms to the cavity mode and on the radiation pressure on the movable mirror can be ignored .
the situation would be different and the center of mass motion important if the optical lattice were generated along the cavity axis by the cavity mode @xcite , or by a standing wave formed by running and reflected from the movable mirror laser beams @xcite .
the motion of the movable mirror is modeled as a quantum mechanical harmonic oscillator of mass @xmath10 and resonant frequency @xmath11 .
the cavity mode is driven by an external laser field which is treated classically in our calculations and is characterized by its frequency @xmath12 and amplitude @xmath13 .
it has become common to consider the laser field as a source of the radiation pressure force on the movable mirror . [ cols="^ " , ] figure [ mirfig5 ] shows the logarithmic negativity @xmath14 for fixed @xmath15 and gradually increasing @xmath16 and @xmath17 . as in the case of two coupled modes
, the logarithmic negativity is determined from the steady - state correlation matrix , which in the present case is a @xmath18 matrix derived by solving an equation that is formally identical with eq .
( [ e61e ] ) , but with the drift matrix @xmath19 given by @xmath20 and with the diffusion matrix @xmath21 $ ] .
it can be seen from fig .
[ mirfig5 ] that entanglement occurs for @xmath22 where it is present for all @xmath17 .
the degree of the entanglement increases with @xmath17 , the strength of the linear mixed process .
note that the entanglement is reduced when @xmath17 is comparable to @xmath23 , and increases as @xmath17 departures from @xmath23 .
we conclude this section that two independent modes may become entangled by a suitable coupling to an intermediate mode .
the entanglement occurs when one of the modes couples to the intermediate mode by the parametric interaction process whereas the other couples by the linear - mixing interaction .
[ sec6 ] finally , we examine parameter ranges in which the predicted coherence and entanglement effects could be observed with the current experiments . as a check on the validity of the limitation to only the @xmath24 exciton
, we evaluate the formula ( [ e20 ] ) for @xmath25 using experimentally realistic parameters of an optical lattice composed of @xmath26rb atoms @xcite . by taking @xmath27 sites , the cavity mode volume @xmath28 [ m@xmath29 , the atomic transition dipole moment @xmath30 [ cm ] of the hyperfine transition 5 @xmath31 5 @xmath32 in an @xmath26rb atom , and the cavity frequency @xmath33 on resonance with the first @xmath24 exciton mode that is comparable to the atomic transition frequency of @xmath34 [ hz ]
, we obtain for the coupling strength of the first @xmath24 exciton , @xmath35 [ hz ] , and for the third @xmath36 exciton , @xmath37 [ hz ] that is one order smaller than the @xmath24 coupling strength .
this simple estimation of the coupling strength shows that @xmath25 decays very fast with @xmath38 .
thus , our theory can describe the single exciton system quite accurately .
the stability condition for the three mode coupling case requires @xmath39 , where @xmath40 and @xmath41 . since @xmath42 [ hz ] and the typical coupling strengths @xmath43 [ hz ]
, the inequality @xmath39 can be satisfied with the realistic experimental parameters .
[ sec7 ] in summary , we have presented an analytical study of coherence and correlation effects produced in a single - mode nano - mechanical cavity containing an optical lattice of regularly trapped atoms . the system considered is equivalent to a three - mode system composed of two polariton modes and one mechanical mode . we have shown that the system is capable of generating a wide class of coherence and correlation effects , ranging from the first - order coherence , the anomalous autocorrelations and anomalous cross correlations between the modes .
we have been particularly interested in the relationship between the generation of entanglement and the first - order coherence in the system .
the results show that the generation of the first - order coherence between two modes of the system is equally effective in destroying entanglement between these modes .
there is no entanglement between the independent polariton modes when both modes are simultaneously coupled to the mechanical mode by the parametric ( squeezing - type ) interaction .
there is no entanglement between the polaritons even if the oscillating mirror is damped by a squeezed vacuum field .
the intermediate mechanical mode effectively creates the first - order coherence between the modes .
finally , we have shown that in order to effectively entangle independent modes , in this system , one of the modes should be coupled to the intermediate mode by a parametric interaction but the other mode should be coupled by the linear - mixing ( beamsplitter - type ) interaction .
this work was supported by the national natural science foundation of china ( grant no .
11074087 ) , , and the natural science foundation of hubei province ( grant no .
2010cda075 ) , the nature science foundation of wuhan city ( grant no .
201150530149 ) , and the national basic research program of china(grant no .
2012cb921602 ) . | the coherence and correlation properties of effective bosonic modes of a nano - mechanical cavity composed of an oscillating mirror and containing an optical lattice of regularly trapped atoms are studied .
the system is modelled as a three - mode system , two orthogonal polariton modes representing the coupled optical lattice and the cavity mode , and one mechanical mode representing the oscillating mirror .
we examine separately the cases of two - mode and three - mode interactions which are distinguished by a suitable tuning of the mechanical mode to the polariton mode frequencies . in the two - mode case
, we find that the occurrence of entanglement in the system is highly sensitive to the presence of the first - order coherence between the modes .
in particular , the creation of the first - order coherence among the polariton and mechanical modes is achieved at the expense of entanglement between them . in the three - mode case , we show that no entanglement is created between the independent polariton modes if both modes are coupled to the mechanical mode by the parametric interaction .
there is no entanglement between the polaritons even if the oscillating mirror is damped by a squeezed vacuum field .
the interaction creates the first - order coherence between the polaritons and the degree of coherence can , in principle , be as large as unity .
this demonstrates that the oscillating mirror can establish the first - order coherence between two independent thermal modes .
a further analysis shows that two independent thermal modes can be made entangled in the system only when one of the modes is coupled to the intermediate mode by a parametric interaction and the other is coupled by a linear - mixing interaction . |
in spinor becs , the mean - field energy is much larger than the spin - exchange interaction .
therefore , we first determine the spatial dependence of the condensate wave function independently of the spin degrees of freedom , and then , we consider the spin state by assuming that the obtained spatial wave function is shared by all spin states . this approximation is called the single - mode approximation . in this section ,
we examine the many - body spin states by using this approximation .
suppose that the coordinate part of the ground - state wave function @xmath128 is independent of the spin state .
then , @xmath128 is the lowest - energy solution of the gp equation @xmath129\phi_0=\mu\phi_0 , \label{mb1}\end{aligned}\ ] ] where @xmath130 is the trapping potential .
let @xmath131 and @xmath132 be the first- and second - lowest eigenvalues of eq .
( [ mb1 ] ) , respectively .
the criterion for the validity of the single - mode assumption is that the first excited energy @xmath133 is much larger than other energy scales : @xmath134 where @xmath135 is the effective volume of the system .
when this condition ( [ mb2 ] ) is satisfied , the field operator @xmath136 @xmath137 may be approximated as @xmath138 , where @xmath139 is the annihilation operator of the bosons that have magnetic quantum number @xmath140 and coordinate wave function @xmath141 . as explained above
, the field operator of a spin-1 bec in the single - mode approximation can be expressed as @xcite @xmath142 for the sake of simplicity , let us consider the case of zero magnetic field ( i.e. , @xmath143 ) . substituting eq .
( [ mb3 ] ) in eq .
( [ v(f=1)2 ] ) and replacing @xmath144 and @xmath145 with @xmath146 and @xmath147 , respectively , we obtain @xmath148 where @xmath149 and @xmath150 ; @xmath141 is determined by eq .
( [ mb2 ] ) subject to the normalization condition @xmath151 substituting the solution of eq .
( [ mb2 ] ) back into ( [ mb4 ] ) , we obtain @xmath152 where @xmath153 . from eq .
( [ operatoridentity(f=1 ) ] ) , we find that the total spin operator @xmath154 satisfies @xmath155 where @xmath156 is the spin - singlet pair operator .
we use eq .
( [ mb8 ] ) to rewrite eq .
( [ mb7 ] ) as @xmath157 . \label{mb10}\end{aligned}\ ] ] because @xmath146 and @xmath158 commute , the eigenvalue problem reduces to finding their simultaneous eigenstates .
let @xmath159 be the state in which there is no spin - singlet pair , i.e. , @xmath160 . using the commutation relation @xmath161 = \frac{2}{3}k \ ; ( 2\hat{n}+2k+1)(\hat{s}^\dagger)^{k-1 } , \label{mb11}\end{aligned}\ ] ] we obtain @xmath162
we then define the @xmath163-pair state @xmath164 : @xmath165 it follows from the commutation relation ( [ mb11 ] ) that @xmath166 next , we consider a state @xmath167 in which @xmath168 atoms form spin - singlet pairs and the remaining @xmath169 atoms reside in the @xmath122 state along the @xmath170 direction @xmath171 . defining @xmath172 the corresponding eigenstate and eigenenergy
are given by @xmath173 \ !
, \label{mb17}\end{aligned}\ ] ] where @xmath174 is the normalization constant . in the absence of an external magnetic field ,
all magnetic sublevels are degenerate .
if we take the quantization axis along the @xmath1-direction , we can specify the eigenstate in terms of the total spin @xmath175 , magnetic quantum number @xmath176 , and number of spin - singlet pairs @xmath163 : @xmath177 where @xmath178 , and @xmath179 is the lowering operator of the magnetic quantum number .
+ next , we consider a situation in which all @xmath175 atoms points in the @xmath1-direction in addition to the @xmath163 spin - singlet pairs , i.e. , @xmath180 in eq .
( [ mb18 ] ) , and seek the number of the atoms that are present in the @xmath122 state .
we introduce the basis state @xmath181 in which the magnetic sublevels @xmath182 are occupied by @xmath183 and @xmath184 atoms , respectively .
then , @xmath185 and @xmath186 substituting @xmath187 and using the fact that @xmath188 commutes with @xmath189 , we have @xmath190 we use eq .
( [ mb11 ] ) to rewrite @xmath191 + ( \hat{s}^\dagger)^k \hat{s } \nonumber \\ & = & \frac{2}{3}k \ ; ( 2\hat{n } + 2k + 1)(\hat{s}^\dagger)^{k-1 } + ( \hat{s}^\dagger)^k \hat{s}. \label{mb22}\end{aligned}\ ] ] noting that @xmath192 and @xmath193 , we have @xmath194 & = & \frac{(2f+2k+3 ) ( 2f+2k+1 ) \cdots\cdot ( 2f+5 ) } { ( 2f+2k+1 ) ( 2f+2k-1 ) \cdots\cdot ( 2f+3 ) } -1 \\ & = & \frac{2k}{2f+3 } = \frac{n - f}{2f+3}.\end{aligned}\ ] ] because @xmath195 and @xmath196 , we obtain @xmath197 when all atoms form spin - singlet pairs , i.e. , @xmath198 , eqs .
( [ mb23])([mb25 ] ) show that all magnetic sublevels are equally populated : @xmath199 that is , the condensate is fragmented in the sense that more than one single - particle state is macroscopically occupied @xcite .
this is in sharp contrast with the mean - field result that predicts that for @xmath200 , the ground state is either polar @xmath201 or antiferromagnetic @xmath202 .
this discrepancy originates from the mean - field assumption that there exists one and only one bec .
however , such an assumption can not be justified when the system possesses certain exact symmetries such as rotational symmetry in the present case .
equation ( [ mb24 ] ) shows that @xmath203 decreases rapidly with an increase in @xmath175 .
in fact , when @xmath204 , we have @xmath205 .
this implies that although mean - field results break down at zero magnetic field , the validity of mean - field theory quickly recovers as the magnetic field increases .
this affords an example of why fragmented becs are fragile against symmetry - breaking perturbations .
+ finally , we note an interesting relationship between a mean - field state and the many - body spin - singlet state .
a mean - field state is one in which all particles occupy the same single - particle state : @xmath206 where @xmath207 is defined in eq .
( [ mb15 ] ) . substituting @xmath208 , and @xmath209 in eq .
( [ mb15 ] ) , we obtain @xmath210 we calculate the equal - weighted superposition state of @xmath211 over all solid angles @xmath212 .
@xmath213 this result shows that the spin - singlet state is the superposition state of a ferromagnetic state @xmath214 over all angles of magnetization .
conversely , the mean - field state @xmath215 may be interpreted as a broken symmetry state of the spin - singlet state with respect to the direction of quantization @xmath216 .
next , we discuss the many - body states of spin-2 becs in the single - mode approximation . substituting @xmath217 in eq .
( [ v(f=2 ) ] ) , we obtain the spin - dependent part of the hamiltonian as @xmath218 where @xmath219 is the spin vector operator , @xmath220 is its @xmath1-component , and @xmath221 are the creation ( @xmath222 ) and annihilation ( @xmath223 ) operators of a spin - singlet pair .
while the operator @xmath222 , when applied to a vacuum , creates a pair of bosons in the spin - singlet state , the pair should not be regarded as a single composite boson because @xmath222 does not satisfy the commutation relations of bosons .
in fact , the operators @xmath224 together with @xmath225 satisfy the @xmath226 commutation relations : @xmath227=\pm\hat{\cal s}_\pm , \ \ [ \hat{\cal s}_+,\hat{\cal s}_-]=-2\hat{\cal s}_z . \label{su(1,1)}\end{aligned}\ ] ] here , the minus sign in the last equation is the only difference from the usual spin commutation relations .
this difference , however , leads to some important consequences .
in particular , the casimir operator @xmath228 , which commutes with @xmath224 and @xmath229 , does not take the usual form of the squared sum of spin components but instead takes the form @xmath230 the eigenstates are classified according to the eigenvalues of the casimir operator ( [ casimir ] ) . because @xmath231 is positive semidefinite , @xmath229 must have a minimum value @xmath232 .
recalling that @xmath233 , @xmath232 can be expressed in terms of a non - negative integer @xmath234 as @xmath235 .
let @xmath236 be the eigenvector corresponding to this minimum eigenvalue ; then , @xmath237 .
it follows that @xmath238 hence , the eigenvalue of @xmath228 is given by @xmath239 with @xmath240 .
the operation of @xmath222 on @xmath236 increases the eigenvalue of @xmath229 by 1 and that of @xmath146 increases by 2 , as observed from the commutation relations ( [ su(1,1 ) ] ) .
therefore , the allowed combinations of eigenvalues @xmath241 and @xmath242 for @xmath228 and @xmath229 , respectively , are given by @xmath243 and @xmath244 where @xmath234 and @xmath245 satisfy @xmath246 and we may therefore interpret @xmath245 as the number of spin - singlet `` pairs '' and @xmath234 the number of remaining bosons . to find the exact energy eigenvalues of hamiltonian ( [ h2 ] ) , we first note that @xmath224 and @xmath229 commute with @xmath247 : @xmath248=0 , \ \
[ \hat{\cal s}_z,\hat{\cal f}]=0.\end{aligned}\ ] ] the energy eigenstates can be classified according to quantum numbers @xmath234 and @xmath245 , total spin @xmath175 , and magnetic quantum number @xmath176 .
we denote the eigenstates as @xmath249 , where @xmath250 labels orthonormal degenerate states with @xmath251 provided in ref .
@xcite . the first commutator in eq .
( [ su(1,1 ) ] ) implies that @xmath224 plays the role of changing the eigenvalue of @xmath229 by @xmath252 by creating or annihilating a spin - singlet pair .
the eigensystem can be constructed by first preparing the eigenstates that do not involve spin - singlet pairs , @xmath253 on it : @xmath254 where @xmath255 , and the orthnormality condition is @xmath256 to write down the eigenstates ( [ eigenstate ] ) explicitly , we need to find the matrix elements of @xmath257 , of which only the nonvanishing ones are @xmath258 finally , the energy eigenvalue for the state @xmath249 is given by @xmath259 + \frac{c_2}{5v^{\rm eff}}n_{\rm s}(n+n_0
+ 3 ) -pf_z , \label{e2}\ ] ] where the relationship @xmath260 is used .
the total spin @xmath175 can , in general , take integer values in the range @xmath261 . however , there are some forbidden values @xcite . that is ,
@xmath262 are not allowed when @xmath263 , and @xmath264 are forbidden when @xmath265 .
we prove this at the end of this subsection .
to gain a physical insight into the nature of the energy eigenstates @xmath249 , it is useful to express them in terms of some building blocks .
in fact , we can express them in terms of one- , two- , and three - boson creation operators .
let @xmath266 be the operator such that when applied to the vacuum state , it creates @xmath10 bosons that have a total spin @xmath267 and magnetic quantum number @xmath268 . while the choice of a complete set of operators for the building blocks is not unique , we choose the following five operators for constructing the eigenstates : @xmath269 = \sqrt{\frac{2}{5}}\hat{\cal s}_+ \\ \hat{a}^{(2)\dagger}_2&= & \frac{1}{\sqrt{14 } } [ 2\sqrt{2}\hat{a}_2^\dagger\hat{a}_0^\dagger -\sqrt{3}(\hat{a}_1^\dagger)^2 ] \\
\hat{a}^{(3)\dagger}_0&= & \frac{1}{\sqrt{210 } } [ \sqrt{2}(\hat{a}_0^\dagger)^3 - 3\sqrt{2}\hat{a}_1\dagger \hat{a}_0^\dagger\hat{a}_{-1}^\dagger + 3\sqrt{3}(\hat{a}_1^\dagger)^2\hat{a}_{-2}^\dagger \nonumber \\ & & + 3\sqrt{3}\hat{a}_2^\dagger(\hat{a}_{-1}^\dagger)^2 -6\sqrt{2}\hat{a}_2^\dagger\hat{a}_0^\dagger \hat{a}_{-2}^\dagger ] \label{def_a30 } \\ \hat{a}^{(3)\dagger}_3&= & \frac{1}{\sqrt{20 } } [ ( \hat{a}_1^\dagger)^3-\sqrt{6}\hat{a}_2^\dagger\hat{a}_1^\dagger \hat{a}_0^\dagger + 2(\hat{a}_2^\dagger)^2\hat{a}_{-1}^\dagger].\end{aligned}\ ] ] note that @xmath270 and @xmath271 vanish identically because of the bose symmetry , and that the operators @xmath272 commute with @xmath273 .
we next consider a set @xmath274 of unnormalized states : @xmath275 with @xmath276 and @xmath277 .
the total number of bosons and the total spin of the state ( [ def_state_unit ] ) are @xmath278 and @xmath279 , respectively . for given @xmath280 and @xmath175 , @xmath281
is uniquely determined by the parity of @xmath175 : @xmath282 it can be shown @xcite that @xmath274 forms a nonorthogonal complete basis set of the subspace @xmath283 in which the magnetic quantum number @xmath176 is equal to the total spin @xmath175 .
it can also be shown @xcite that the energy eigenstates can be represented as @xmath284 where @xmath276 , @xmath277 , and @xmath285 .
these parameters are related to @xmath286 as @xmath287 and the corresponding eigenenergy is given by eq .
( [ e2 ] ) .
note that the states defined in ( [ eigenstates ] ) are unnormalized , and the states having the same energy ( i.e. , those belonging to the same set of parameter values @xmath286 ) are nonorthogonal .
it might be tempting to envisage a physical picture that the system , as in the case of @xmath288he , to be made up of @xmath289 composite bosons whose creation operator is given by @xmath272
. however , this is an oversimplification because the operator @xmath290 does not obey the boson commutation relation .
moreover , the projection operator @xmath291 in ( [ eigenstates ] ) imposes many - body spin correlations such that the spin correlation between _ any _ two bosons must have a vanishing spin - singlet component .
note that two bosons with independently fluctuating spins generally have a nonzero overlap with the spin - singlet state .
the many - body spin correlations of the energy eigenstates are thus far more complicated than what a naive picture of composite bosons suggests . on the other hand ,
as long as quantities such as the number of bosons , magnetization , and energy are concerned , the above simplified picture is quite helpful .
as an illustration , we provide an explanation for the existence of forbidden values for the total spin @xmath175 .
for example , to construct a state with @xmath198 or @xmath292 , composite particles with total spin 2 must be avoided , namely , @xmath293 .
then , we have @xmath294 , implying that @xmath198 or @xmath292 is only possible when @xmath295 . for a state with @xmath296 or @xmath297
, we have @xmath298 and @xmath299 , implying that @xmath300 .
the above simplified picture is also helpful when we consider the magnetic response discussed below . here
, we consider how the ground state and the magnetization @xmath176 respond to the applied magnetic field @xmath301 . from eq .
( [ e2 ] ) , we see that the minimum energy states always satisfy @xmath302 when @xmath303 .
the problem thus reduces to minimizing the function @xmath304 + \frac{c_2}{5v^{\rm eff}}n_{\rm s}(2n-2n_{\rm s}+3 ) -pf_z . \label{e3}\ ] ] here , we only consider the case of @xmath305 .
detailed investigations of this problem can be found in ref .
@xcite . in this case
, it is convenient to introduce a new parameter @xmath306 and consider the energy as a function of @xmath176 and @xmath307 : @xmath308 ^ 2 \nonumber \\ & & -\frac{c_2}{40v^{\rm eff}}l(l+2f+6)+{\rm const}. \label{e_fz_l}\end{aligned}\ ] ] while the averaged slope @xmath309 coincides with that in mft , the offset term @xmath310 in eq .
( [ e_fz_l ] ) makes a qualitative distinction from mft , that is , the onset of the magnetization displaces from @xmath311 to @xmath312 .
note that the slope @xmath313 and the offset @xmath314 are determined by independent parameters . a typical behavior of the magnetic response when @xmath315 is shown in fig . [ fig : spin2meissner ] .
we now calculate the zeeman - level populations of the ground states for @xmath305 . in mft ,
the lowest - energy states for @xmath305 have vanishing population in the @xmath316 levels .
in contrast , the exact ground states derived in the preceding subsection , @xmath317 with @xmath318 and @xmath277 , have nonzero populations in the @xmath316 levels .
the exact forms for the averaged population @xmath319 are calculated as follows .
the above ground states have the form @xmath320 , with @xmath236 being a state with a fixed number ( @xmath321 ) of bosons satisfying @xmath237 . the average zeeman population for the ground states , @xmath322 is then simply related to the average zeeman populations for the state @xmath236 as @xmath323 where @xmath324 .
this formula implies that when @xmath325 , the zeeman populations of the ground states are sensitive to the form of @xmath236 . with this formula , it is a straightforward task to calculate the average zeeman - level populations for four types of ground states , @xmath326 with @xmath318 and @xmath277 .
a striking feature appears in the leading terms under the condition @xmath327 .
the results are summarized as @xmath328 and @xmath329 these results show that the population of each magnetic sublevel depends very sensitively on the spin correlations . as a consequence ,
a minor change in the magnetization might lead to a major change in the population .
such dramatic changes originate from bosonic stimulations caused by the term @xmath330 in @xmath331 or the term @xmath332 in @xmath333 .
the possible ground states of spinor becs can be inferred based on symmetry arguments without solving the gpes .
the spin-2 bec has five internal degrees of freedom corresponding to magnetic quantum numbers @xmath334 .
let @xmath335 s be their amplitudes . under rotation of the coordinate system , they must transform so as to guarantee that the linear combination @xmath336 is scalar , where @xmath337 is the spherical harmonic function of rank 2 with @xmath55 and @xmath56 being the polar and azimuthal angles , respectively , in the spherical coordinate system .
let @xmath338 be the unit vector in the @xmath216 direction . in terms of the unit vector @xmath339
, we can rewrite eq .
( [ rank2tensor ] ) as @xmath340 where superscript @xmath341 denotes the transpose , and @xmath342 the order parameter is thus characterized by a @xmath343 traceless symmetric matrix tr@xmath344 with unit normalization @xmath345 let us now consider the eigenvalues of this order parameter matrix : @xmath346 where @xmath347 is added in front of the matrix for the sake of convenience , and @xmath348 and @xmath341 are the amplitudes of the spin - singlet pair and spin - singlet trio , respectively : @xmath349 substituting @xmath350 and @xmath351 , eq .
( [ determinant ] ) reduces to @xmath352 and hence , the solutions are @xmath353 where @xmath354 and @xmath355 we can infer from this that the spin - singlet pair and the spin - singlet trio of atoms form building blocks of the ground state of the spin-2 bec .
it is noteworthy that this fact arises solely from the fact that the order parameter matrix of the spin-2 bec is traceless symmetric .
topological excitations can be spontaneously created when a system undergoes a phase transition in a finite time .
the basic concept is that after the phase transition domains of the new phase emerge at causally disconnected places , and therefore , the order parameters of the new phase at different locations are not correlated .
when the order parameters grow to overlap , they may be able to adjust dynamically so that they are connected smoothly ; if they can not , singularities should develop in the order parameter space , giving rise to topological excitations .
such a scenario of topological defect formation was first discussed by kibble @xcite in the context of cosmic - string and monopole formations in the early universe , and an experimental test for the kibble scenario in condensed matter systems was proposed by zurek @xcite ( see ref .
@xcite for a review ) .
bose - einstein condensates of dilute atomic vapor offer an ideal opportunity for studying the kibble - zurek mechanism because the temperature , strength of interaction , and external parameters such as magnetic field and trapping potentials can be changed in a time shorter than the characteristic time scale of topological defect formation .
in fact , scalar vortices @xcite and spin vortices @xcite have been observed to emerge spontaneously upon both temperature and magnetic - field quenches .
topological defect formation via the kibble - zurek mechanism can occur in a ferromagnetic phase of a spinor bec , where the phase transition is triggered by a sudden change in a magnetic field ( magnetic field quenching ) @xcite .
the ground - state phase of a spin-1 ferromagnetic bec is shown in fig .
[ fig : spin-1pd ] , where @xmath301 and @xmath356 are the coefficients of the linear and quadratic zeeman effects , respectively .
when we consider a type of magnetic - field quench for which the total magnetization of the system is conserved , the quench occurs along a constant-@xmath301 line of the phase diagram . as in the experiment described in ref .
@xcite , we consider the case of @xmath311 that implies that the total magnetization of the system is kept to be zero .
then , the state of the system can be in one of the following two phases . when @xmath356 is larger than the critical value @xmath357 , with @xmath10 being the number density , the ground state is the polar phase @xmath358 which is nonmagnetic . when @xmath359 , the ground state is the broken - axisymmetry ( ba ) phase in which the system develops transverse magnetization . the ground state for @xmath359
is the broken - antisymmetry ( ba ) phase with the order parameter given by @xmath360 where @xmath361 .
the ba phase has transverse magnetization @xmath362 which breaks the @xmath363 symmetry corresponding to the direction of the transverse magnetization . by rapidly quenching the magnetic field from above to below @xmath364 ,
local transverse magnetization develops in random directions , and thus , spin vortices are spontaneously created . to analyze such a situation ,
let us assume that the initial state is the polar phase @xmath365 and expand the field operator as @xmath366 substituting this into the hamiltonian and keeping the terms up to the second order in @xmath367 with @xmath368 , we obtain the bogoliubov hamiltonian for the polar phase .
we are interested in how the modes @xmath369 grow with time .
the heisenberg equations of motion for the modes are @xcite @xmath370 with the solutions given by @xmath371 where @xmath372 gives the bogoliubov spectrum .
if @xmath373 has an imaginary part , the corresponding mode is dynamically unstable and it grows exponentially . because @xmath374 and @xmath375 for spin-1 @xmath376rb atoms , exponential growth occurs for @xmath359 , in agreement with the phase boundary .
we can use the solution ( [ solution ] ) to investigate how the transverse magnetization develops with time . because @xmath377 , the magnetization operator @xmath378 is written as @xmath379 it follows that the correlation function of the transverse magnetization is given by @xmath380 the exponential terms describe dynamical instabilities of which the dominant term is the one for which the imaginary part of @xmath373 is maximal .
let @xmath381 be the `` most unstable '' wave number for which @xmath382 is maximal .
it follows from eq .
( [ e_k ] ) that @xmath383 for @xmath384 and @xmath385 for @xmath386 .
thus , for @xmath386 , the typical size of magnetic domains is @xmath387 ; for @xmath384 , it is determined by the width of the distribution of im@xmath373 . in any case
, the size of the spin domains grows as the final value of @xmath356 increases , as experimentally observed in ref .
the time scale @xmath388 for the magnetization is given by @xmath389 which is @xmath390 for @xmath384 and @xmath391 for @xmath386 .
next , we consider the case in which the magnetic field is quenched linearly in a finite time @xmath392 : @xmath393 in this case , the bogoliubov excitation energy ( [ e_k ] ) also varies with time and the spin correlation function is estimated to be @xmath394 \nonumber \\ & \propto & \exp\left [ f(t)-|{\bm r}-{\bm r}'|^2/\xi_{\rm q}^2 \right],\end{aligned}\ ] ] where @xmath395 is the correlation length and @xmath396 determines the growth rate of magnetization . in two dimensions , we have @xcite @xmath397^\frac{1}{4 } \label{tau_q}\end{aligned}\ ] ] and @xmath398 . \label{f(t)}\end{aligned}\ ] ] the time scale for magnetization is determined by @xmath399 . solving this equation by assuming that @xmath400
, we obtain @xmath401 substituting this into eq .
( [ tau_q ] ) , we obtain @xmath402 the same power law was obtained in ref .
@xcite .
the topological defect formation caused by a quantum phase transition can be best illustrated by numerical simulations .
figure [ fig : kzmag ] shows the time evolution following an instantaneous quench of the magnetic field from @xmath403 to @xmath404 ( fig .
[ fig : kzmag](a ) ) and @xmath405 . here
, we assume that the system is a two - dimensional disk with a hard wall at radius 100 @xmath406 m and the potential is assumed to be flat inside the wall .
the initial state is considered to be a stationary state of the gp equations and the following quench dynamics is obtained by using the time - dependent gp equations . to trigger the dynamical instabilities that cause defect formation , we assume small fluctuations in the initial amplitudes of the @xmath369 components according to @xmath407 where @xmath408 and @xmath409 are random variables whose amplitudes @xmath410 follow the normal distribution @xmath411 .
figure [ fig : kzmag ] shows snapshots of the amplitude @xmath412 and the phase arg@xmath413 of the transverse magnetization after the quench .
we observe that many holes emerge spontaneously after 100 ms .
these holes represent topological excitations called polar - core vortices in which the @xmath369 components have vortices with the cores filled by the @xmath122 components . by counting the number of vortices in the final state , we can determine the number of spin vortices .
alternatively , we can calculate the same number the spin winding number @xmath414from the algebraic relation @xmath415 where @xmath416 is a circle of radius @xmath15 and @xmath414 is a count of the number of rotations of the spin vector in the @xmath127 plane along @xmath416 .
figure [ fig : kzrdep ] shows the @xmath15 dependence of the ensemble average of @xmath417 .
we note that @xmath418 is proportional to @xmath15 for large @xmath15 in agreement with the kz theory , whereas it is proportional to @xmath419 for small @xmath15 .
if spin vortices can be generated randomly , @xmath418 should be proportional to the area of the system and hence to @xmath419 .
the kz scaling law @xmath420 that we observe for large @xmath15 arises from the fact that the number of vortices matches that of anti - vortices inside the disk due to spin conservation . in other words , for small @xmath356 ,
magnetic domains are aligned in such a manner as to cancel the local spin when averaged over the spin correlation length .
thus , over a greater length scale , the magnetic domains can be created independently .
this is the underlying physics that makes the spin conservation compatible with the kz postulate of independent defect creation at a long distance and the kz scaling law valid for 2d systems .
it had been conjectured from the work of peierls and landau in the 1930s @xcite that two - dimensional systems with continuous order parameters could not have a conventional long - range order , and this was confirmed by mermin and wagner @xcite and by hohenberg @xcite .
long wavelength fluctuations are more important in one and two dimensions than in three dimensions and deviations from the mean - field theory are much greater .
however , there still exists a type of quasi - long - range order in the low - temperature phase in two - dimensional systems that is characterized by a power - law decay of the order parameter correlation function .
such a phase transition is known as the berezinskii - kosterlitz - thouless ( bkt ) transition . in this subsection
, we first explain the bkt transition for a single - component bose gas , and then , we extend it to a spinor gas .
we start from a ginzburg - landau energy functional @xmath421 = \int d^2 r \left [ \frac{j}{2 } |\nabla \psi|^2+\frac{\alpha}{2 } |\psi|^2 + \frac{\beta}{4 } |\psi|^4 \right ] , \label{eq:2d_glenergy}\end{aligned}\ ] ] which gives an equilibrium probability for @xmath422 as @xmath423 \propto \exp[-\frac{h}{k_{\rm b}t}]$ ] . here
, @xmath424 and @xmath408 changes its sign from positive to negative as the temperature decreases from above to below the critical temperature @xmath425 , so that the equilibrium value of @xmath422 changes from @xmath426 at @xmath427 to @xmath428 at @xmath429 . in three dimensions ,
the one - body correlation function @xmath430 remains finite at large distance below @xmath425 , exhibiting a long - range order .
in constrast , above @xmath431 , the correlation function decays exponentially : @xmath432 where @xmath433 except in the immediate vicinity of @xmath425 ( see e.g. ref .
@xcite ) . in this case , the order is short - ranged . in two dimensions ,
the effect of fluctuations is much more drastic .
we first assume that @xmath434 takes its equilibrium value everywhere and consider only phase fluctuations . substituting @xmath435 in eq .
, we have @xmath436 where @xmath437 and @xmath438 is the fourier component of @xmath439 satisfying @xmath440 .
because the gradient of @xmath56 induces a superfluid current as @xmath441 , the coefficient @xmath442 is related to the superfluid density @xmath443 as @xmath444 so that @xmath445 is rewritten as @xmath446 because eq .
is quadratic in @xmath438 , we can calculate the correlation function as @xmath447.\end{aligned}\ ] ] in a two - dimensional system , the summation of @xmath339 is replaced by an integral as @xmath448\\ & = \frac{k_{\rm b}t}{2\pi k_0 } \int dk \frac{1-j_0(kr)}{k},\end{aligned}\ ] ] where @xmath449 is the zeroth bessel function . for @xmath450 , we can ignore the rapidly oscillating function @xmath451 , whereas the contribution from @xmath452 is small .
we also introduce a cutoff wavelength @xmath453 corresponding to the coherence length , and neglect spatial structures smaller than @xmath453 .
then , the correlation function is calculated as @xmath454 \propto \left(\frac{r}{\xi}\right)^{-\eta},\end{aligned}\ ] ] where @xmath455 thus , phase fluctuations destroy the true long - range order ; however , the quasi long - range order characterized by a power - law behavior remains .
next , we consider the effect of fluctuations on the amplitude @xmath434 .
a crucial effect arises when @xmath434 passes through zero .
if @xmath456 is finite everywhere in a singly connected region , the phase @xmath56 of the order parameter is defined as a single - valued function . on the other hand ,
if @xmath426 at @xmath457 , @xmath56 can be a multi - valued function .
the points @xmath458 correspond to vortices around which @xmath56 varies by an integral multiple of @xmath459 . the phase transition for the appearance of a quasi - long - range order can be captured by a simple discussion of vortex nucleation . when a single vortex passes between two distinct points , the relative phase between these two points changes by @xmath460 , implying that free vortices destroy the phase coherence .
hence , the quasi - long - range order is destroyed when free vortices are thermally excited .
the free energy for a single vortex can be calculated as follows .
the circulation of the superfluid velocity around the vortex is quantized as @xmath461 .
assuming axisymmetry around the vortex , we have @xmath462 , where @xmath463 is the polar coordinate around the vortex and @xmath82 is a unit vector in the direction of @xmath83 .
then , the energy cost required to create a single vortex is given by @xmath464 where @xmath15 is the radius of the system ; @xmath453 , the radius of the vortex core ; and @xmath465 , the vortex core energy arising from the region of @xmath466 .
on the other hand , because the vortex can freely move in the area of @xmath419 , the entropy is given by @xmath467 .
then , we obtain the free energy of a free vortex as @xmath468 one can immediately see from this free energy form that the phase transition occurs at @xmath469 above which free vortices are generated and destroy the phase coherence . from eqs . and , the exponent @xmath470 must satisfy @xmath471 .
a more careful argument requires the interaction between vortices to be considered .
if there exists a pair of a vortex ( charge 1 ) and an anti - vortex ( charge @xmath472 ) at distance @xmath473 , the velocity fields generated by these vortices at a distance larger than @xmath473 cancel each other out . as a result
, the vortex pair has a finite energy of @xmath474 which indicates that a vortex and an anti - vortex attract each other and form a pair at low temperature . because vortex pairs do not destroy phase coherence ( see fig .
[ fig:2d_vortex_pair ] ) , the superfluid phase transition can be interpreted as a dissociation of vortex pairs .
if there exist many vortex pairs , the attractive interaction between a vortex and an anti - vortex is weakened by the screening effect .
kosterlitz and thouless @xcite analyzed this phase transition by mapping the system to a two - dimensional coulomb gas of unit charge @xmath475 .
they introduced a `` dielectric constant '' @xmath476 that renormalizes @xmath442 to @xmath477 , and constructed the renormalization - group flow equations .
their main result is that the superfluid density jumps at the transition temperature that satisfies the universal relation : @xmath478 this universal behavior in superfluid helium was observed by bishop and reppy @xcite . with regard to cold atom systems , the bkt transition of a single - component bose gas was observed in ref .
@xcite we now consider the case of a spinor .
as discussed above , vortices play a crucial role in the superfluid phase transition in two - dimensional systems .
when a system acquires spin degrees of freedom , the bkt phase transition becomes highly nontrivial because the nature of the vortices depends on the order parameter manifold , as discussed in sec .
[ sec : topology ] .
it is well known that though the xy spin model exhibits a bkt phase transition where the order parameter manifold u(1 ) is the same as that of a single - component bose gas , the bkt transition does not occur for the heisenberg spin model at finite temperature @xcite .
this is because the order parameter manifold for the hisenberg model , which is isomorphic to @xmath46 , does not accommodate a topologically stable vortex because @xmath479 . as discussed in sec .
[ sec : topology ] , the order parameter manifold for the polar phase of a spin-1 bec can accommodate half - quantum vortices , and therefore , the bkt transition occurs due to the binding of half - quantum vortex pairs . on the other hand ,
if the condensate accommodates several types of vortices , the bkt transition is expected to occur in several stages . here
, we discuss two bkt transitions of ferromagnetic spinor gases that occur under an external magnetic field .
for simplicity , we neglect the trapping potential and assume an infinite system . the energy functional for a spin-1 spinor gas under an external magnetic field @xmath480 in the @xmath1-direction is given by @xmath481,\end{aligned}\ ] ] where @xmath482 and @xmath483 are the number density and spin density , respectively , and @xmath356 is the quadratic zeeman energy proportional to @xmath484 . for spin-1 becs , @xmath485 .
the linear zeeman term simply causes larmor precession due to the spin conservation , and therefore , it is omitted by going onto a rotating frame of reference in the spin space .
we consider the ferromagnetic interaction , @xmath374 . at zero temperature ,
the ground state phase is the same as that we have obtained for a three - dimensional system in sec .
[ sec : meanfield ] . here
, we assume that the total longitudinal magnetization is conserved to be zero .
then , the order parameter for @xmath486 is the polar phase whose order parameter is given by @xmath487 whereas for @xmath359 , the broken - axisymmetry phase appears : @xmath488 where @xmath489 is the absolute square of the order parameter at zero temperature and @xmath490 . for @xmath403 , there exists only one type of vortex ; a conventional phase vortex .
therefore , the bkt transition for @xmath403 is the same as that in single - component gases .
unbinding of phase vortex pairs destroys the coherence of the u(1 ) phase that appears in the correlation function of the @xmath122 component : @xmath491 where @xmath470 is given by eq . with renormalized @xmath492 instead of @xmath493 .
the universal jump is given by eq . .
on the other hand , for @xmath359 , there exist two types of vortices , i.e. , a phase vortex and a spin vortex .
the phase vortex is the same as that in the polar state , around which the u(1 ) phase @xmath56 changes by integer multiples of @xmath459 .
the spin vortex is defined as a vortex around which the direction of transverse magnetization ( i.e. , @xmath408 ) varies by integer multiple of @xmath459 . here
, the transverse magnetization appears at @xmath359 and it is given by @xmath494 the effects of these vortices are independent because the energy for these fluctuations are decoupled as @xmath495,\end{aligned}\ ] ] where @xmath496 similar to the discussion for eq .
, a rough estimation for the transition temperature for the appearance of a free phase vortex and a free spin vortex are given by @xmath497 respectively .
hence , as the temperature decreases , phase vortex pairs first bind and then spin vortices form pairs at lower temperature .
more proper treatment following the kosterlitz - thouless s renormalization method gives the relation between the universal jump and the transition temperature : @xmath498 the phase transitions related to the binding of the phase and spin vortices can be observed as an appearance of the quasi - long - range order in the correlation function of @xmath122 and @xmath499 components , respectively , which are given by @xmath500 with @xmath501 the binding of spin vortices also contributes to the ferromagnetic long - range order @xmath502 figure [ fig:2d_phasediagram ] summarizes the above results .
depending on the temperature and the external magnetic field , there exist three ordered phases : ( i ) quasi - long - range order arises only in the @xmath122 component , where the gas has no local magnetization ; ( ii ) a quasi - long - range order arises only in the @xmath122 component , where the gas is locally magnetized , but with no long - range order of magnetization ; ( iii ) a quasi - long - range order appears in every @xmath140 component , where the transverse magnetization also exhibits the quasi - long - range order . here , we discuss the phase diagram at low magnetic fields . in the above system ,
spin vortices are stable because the quadratic zeeman effect suppresses the fluctuation of longitudinal magnetization .
as the external field decreases , however , the contribution of such fluctuations becomes larger and induces coupling between the spin and the phase vortices .
then , both transition temperatures are expected to be suppressed . at zero magnetic field
, the quasi - long - range order does not appear at finite temperature as in the case of the hisenberg spin system .
understanding the detailed behavior in the low - magnetic - field region requires further investigation .
in the present paper , we have reviewed the basic knowledge concerning spinor bose - einstein condensates ( becs ) that has been accumulated thus far .
the fundamental characteristics of spinor becs are rotational invariance , coupling between the spin and the gauge degrees of freedom , and magnetism arising from the magnetic moment of the spin .
the rotational invariance and gauge invariance alone uniquely determine the microscopic hamiltonian of the spinor bec , as discussed in sec .
[ sec : generaltheory ] . the mean - field theory described in sec .
[ sec : meanfield ] discribes the ground - state phases and the dynamics of spinor becs . in a scalar bec ,
the measurements of the collective modes are found to agree with the theoretical values with an accuracy better than 1% ( see ref .
@xcite for a comprehensive review ) . in the near future ,
experimental tasks to be carried out include investigating the extent to which the predictions of the bogoliubov theory described in sec . [ sec : bogoliubov ] can be verified experimentally .
the topological excitations described in sec . [ sec : topology ] are of cross - disciplinary interest ; in particular , the non - abelian nature of vortices in the nematic and cyclic phases of spin-2 becs will have fundamental implications on quantum turbulence and other subfields of physics .
the topological excitations are usually considered within a given order parameter manifold .
however , as exemplified in fig .
[ fig : knottime ] in sec .
[ sec : skyrmions ] , the order - parameter manifold can change dynamically from one to another , and during a transition , the topological charge is no longer conserved .
it is an interesting theoretical question to investigate whether , if any , a more general topologically invariant quantity can be defined in such a general situation .
the relative homotopy theory described in sec .
[ sec : relativehomotopy ] may be used as a guiding theoretical framework .
a closely related subject is vortex nucleation dynamics and vortex lattice formation .
vortices are singularities of the order parameter . in a scalar bec ,
the only possible singularity is the density hole in the condensate .
however , the spinor bec has several different phases and the choice of creating a density hole is energetically most costly .
thus , in spinor becs , the vortex core is filled by a state that does not belong to the order parameter manifold of the condensate . as discussed in sec .
[ sec : vortices ] , the spin-1 polar bec has a ferromagnetic vortex core , and the spin-1 ferromagnetic bec can host a polar - core vortex . for spin-2 and higher - spin becs , more than two possible phases can fill the vortex core .
because each phase has its own internal spin structure , an interesting question arises as to how the order parameter of the bec smoothly connects the spin structure of the vortex core @xcite .
the many - body quantum correlations of spinor becs described in sec . [ sec : manybodytheory ] are fragile against an external magnetic field because the bosonic stimulation favors the mean field .
the spinor - dipolar phenomena described in sec .
[ sec : dipole_spinor ] are also vulnerable against a magnetic field because the larmor precession averages out the part of the dipole interaction that couples the different total spin components .
it is noteworthy that both many - body spin correlations and spinor - dipolar effects become significant as the external magnetic field is reduced down to approximately 10 @xmath406 g .
we have good reasons to expect that further symmetry breaking occurs in such an environment of ultralow magnetic field . from the viewpoint of fundamental physics , the topological defect formation based on the kibble - zurek mechanism discussed in sec .
[ sec : kz ] merits further experimental and theoretical study . in particular , this effect combined with the spinor berezinskii - kosterlitz - thouless transition in two dimensions described in sec .
[ sec : bkt ] raises many interesting questions .
in fact , the gauge , spin , and dipole - dipole interactions set very different energy scales that , we can conceive , lead to topological phase transitions at different energy and temperature scales .
all the discussions presented in this review are restricted to absolute zero .
however , experiments on spinor becs have been carried out at temperatures much higher than the typical energy scale of the spin - exchange interaction .
nevertheless , once the system undergoes bose - einstein condensation , all the condensed particles respond to an external perturbation in exactly the same manner . when the number of condensate particles is of the order of @xmath503 , the collective energy for the dipole interaction amounts to 1 @xmath406k for @xmath504rb and 100 @xmath406k for @xmath505cr .
therefore , for temperatures below 0.1 @xmath406k , the dynamics of the system can be understood , at least qualitatively , based on the zero - temperature theory .
nonetheless , it is likely that the precise measurements of collective modes are sensitive to finite - temperature effects .
moreover , when the long - time dynamics is considered , the thermalization process is expected to play a crucial role in determining the magnetic structure of the system .
in fact , very recent experiments on the @xmath504rb @xmath506 bec show evidence of magnetic ordering @xcite as a consequence of thermalization . to fully understand such phenomena and to explore new types of temperature - quenched symmetry - breaking phase transitions
, it is obvious that we need to develop theoretical tools to investigate the dynamics of spinor - dipolar physics at finite temperatures .
we acknowledge the fruitful collaborations with hiroki saito and michikazu kobayashi .
mu acknowledges the aspen center for physics , where part of this work was carried out . | an overview of the physics of spinor and dipolar bose - einstein condensates ( becs ) is presented .
mean - field ground states , bogoliubov spectra , and many - body ground and excited states of spinor becs are discussed .
properties of spin - polarized dipolar becs and those of spinor - dipolar becs are reviewed .
some of the unique features of the vortices in spinor becs , such as fractional vortices and non - abelian vortices , are delineated .
the symmetry of the order parameter is classified using group theory , and various topological excitations are investigated based on homotopy theory .
some of the more recent developments such as the kibble - zurek mechanism of defect formation and the berezinskii - kosterlitz - thouless transition in a spinor bec are also discussed .
spinor bec , dipolar bec , vortices , topological excitations , low - dimensional systems .table of symbols [ cols="<,<",options="header " , ] to fully understand broken - symmetry systems , it is important to take into account not only the global boundary conditions but also various constraints imposed on the order - parameter field .
relative homotopy groups provide a useful tool for characterizing effects when the order - parameter manifold changes its character on a surface or when it is constrained by boundary conditions .
relative homotopy groups can also be used to discuss properties of a defect that is itself regarded as a singularity in the order - parameter field in ordinary homotopy theory .
finally , the relative homotopy group is useful to calculate the absolute homotopy group via the exact sequence of homomorphisms . as an illustrative example , let us again consider a ferromagnetic bec .
suppose that the system is contained in a cylinder of radius @xmath0 and impose a boundary condition that the spin points in the positive @xmath1-direction at @xmath2 and outward or downward on the wall at @xmath3 , where @xmath4 is the radial coordinate .
then , the spin texture would look like @xmath5 \frac{e^{i\varphi}}{\sqrt{2 } } \sin \beta(r ) \\[1 mm ] e^{2i\varphi } \sin^2 \frac{\beta(r)}{2 } \end{pmatrix}. \label{m - h}\end{aligned}\ ] ] because the direction of spin is @xmath6 , the above boundary conditions are met if we choose @xmath7 and @xmath8 or @xmath9 .
the topological objects corresponding to @xmath8 and @xmath9 are referred to as the mermin - ho vortex and anderson - toulouse vortex , respectively .
it can be shown that the circulation at the wall is 1 for the mermin - ho vortex and 2 for the anderson - toulouse vortex .
they are good examples of relative homotopy because both of them are not stable in free space and they can be stabilized only if the boundary conditions on the wall are imposed by some means . in general , the @xmath10th relative homotopy group @xmath11 consists of the elements of @xmath12 from which those of @xmath13 are subtracted .
we consider an @xmath10-sphere in @xmath14 and let it expand into @xmath15 by continuous deformation .
such an expansion may be regarded as a one - to - one mapping of @xmath13 into a subgroup of @xmath12 .
let im@xmath16 be the image of this homomorphism .
then , by the definition given above , the relative homotopy @xmath11 is a quotient group of @xmath17 in @xmath12 : @xmath18 the relative first homotopy group @xmath19 can cope with planar defects with nontrivial boundaries .
a well - known example is a planar soliton ( maki domain wall ) in the a phase of superfluid helium-3 , where the order - parameter manifold is @xmath20 within a dipole healing length from the domain wall and @xmath21 outside this layer due to dipole locking .
the domain wall may be regarded as a linear defect as seen from the edge of the wall .
we take the axis perpendicular to the domain wall as the @xmath1-axis .
then , the order - parameter field defines a mapping from the @xmath1-axis into the contour in @xmath15 whose end points belong to @xmath14 . because @xmath22 , such contours are also classified as @xmath23 , of which the nontrivial one corresponds to the maki domain wall .
the relative second homotopy group @xmath19 allows the stabilization of a point object subject to a boundary condition on a sphere enclosing it .
another example is the boojum a point defect that appears in the a phase of superfluid helium-3 when it is contained in a hard - wall sphere . because @xmath24 , the bulk superfluid a phase does not support a topologically stable point defect .
however , because the @xmath25 vector must be perpendicular to the surface , a point singularity for the @xmath25-vector appears on the surface that is connected to a nonsingular vortex texture in the bulk .
the line defects are characterized by the first homotopy group ( or the fundamental group ) @xmath26 . for the case of _
s_-wave superconductors , liquid helium-4 , and spin - polarized or spinless gaseous becs , the order parameter is scalar and the order - parameter manifold is @xmath27 .
the fundamental group is therefore the additive group of integers : @xmath28 for the case of a spin-1 bec , the ground state can be ferromagnetic or polar .
the order - parameter manifold of the ferromagnetic phase is so(3 ) and the fundamental group is @xmath29 , the two - element group .
the spin configuration that corresponds to class 0 is continuously transformable to a uniform configuration , whereas the spin configuration that corresponds to class 1 describes a singly quantized vortex that is topologically stable . because 1 + 1 = 0 in @xmath30 , the coalescence of two singly quantized vortices is homotopic to a uniform configuration .
conversely , there is a configuration called the anderson - toulouse vortex that is nonsingular at the origin but has circulation of two away from the origin .
next , let us consider the polar phase of a spin-1 bec .
the order - parameter manifold is @xmath31/({\mathbb z}_2)_{{\bm f},\phi}$ ] . as we have discussed in sec .
[ sec : sym_opm ] , the @xmath32 symmetry arises from the fact that the order parameter given by eq .
is invariant under the simultaneous transformation of @xmath33 .
because @xmath34 is not simply connected , to calculate @xmath35 , we lift @xmath36 to @xmath37 and use eq . .
then , the points that are identical to @xmath38 in @xmath39 are @xmath40 for even @xmath10 and @xmath41 for odd @xmath10 , and those points constitute an additive group of integers @xmath10 .
hence , we have @xmath42 as in the case of a scalar bec , vortices in the polar phase are classified by integers .
however , due to the @xmath32 discrete symmetry , the minimum unit of circulation is one - half of the usual value of @xmath43 , as discussed in sec .
[ sec : vortex_spin1polar ] . in a similar manner
, we may expect that the uniaxial , biaxial , and cyclic phases in spin-2 spinor becs can support various types of vortices due to the discrete symmetry , as discussed in sec .
[ sec : vortices ] .
some of them have fractional circulations , and some others have no mass circulation .
the topological charge can be expressed algebraically as a functional of the order - parameter field . the topological charge corresponding to @xmath44
can be expressed as a line integral of the velocity field .
point defects are characterized by the second homotopy group . because @xmath45 , scalar becs have no topologically stable point defects .
when the order - parameter manifold is @xmath46 , it can host monopoles because @xmath47 . here
, we consider the situation shown in fig .
[ fig : homotopy3 ] , where @xmath48 is a mapping from @xmath46 to @xmath46 .
the topological charge @xmath49 of the point defect can be calculated as follows . by definition
, @xmath49 gives the number of times the vector @xmath48 shown in fig .
[ fig : homotopy3 ] wraps @xmath46 . expressing the components of this vector in spherical coordinate as @xmath50 , @xmath51 , and @xmath52
, we can calculate @xmath49 as @xmath53 where @xmath54 , with @xmath55 and @xmath56 are the spherical coordinates of @xmath57 , and the last term on the right - hand side is the jacobian of the transformation of the coordinates .
the fact that the integrand is the jacobian explicitly indicates that @xmath49 gives the degree of mapping from @xmath46 in real space to @xmath46 in order - parameter space . the right - hand side of eq .
( [ n_2 ] ) can be directly expressed in terms of @xmath58 as @xmath59 where @xmath60 we can use gauss law to rewrite the right - hand side of eq .
( [ n_22 ] ) as a volume integral : @xmath61 where @xmath62 gives the density distribution of point singularities . + * t - hooft - polyakov monopole ( hedgehog ) * next , we consider the polar phase .
the second homotopy group is given by @xmath63 , and therefore , it can accommodate point defects . to investigate their nature , we write the order parameter in the following form : @xmath64 by setting @xmath65 , we obtain the spherical monopole called t - hooft - polyakov monopole or hedgehog : @xmath66 it follows from eq . that the topological charge of the hedgehog is @xmath67 .
moreover , @xmath68 defined in eq .
is related to the circulation of spin currents given by eq . as @xmath69 which implies that the surface integral of the spin circulation is quantized in terms of @xmath70
; @xmath71 + * alice ring * the monopole in the spin-1 polar phase is energetically unstable against deformation into an alice ring @xcite .
the alice ring is a combined object of two defects characterized by @xmath72 and @xmath73 ( see fig .
[ fig : homotopy7 ] ) .
viewed far from the origin , it appears to be a point defect ( @xmath67 ) ; however , along the ring @xmath74 on which the order parameter is singular , it appears to be a continuous distribution of line defects ( @xmath75 ) .
this topological object is called the alice ring , and it has been predicted to be realized in an optically trapped spin-1 @xmath76na bec @xcite .
+ * dirac monopole * the magnetic monopole was originally envisaged as a magnetic analogue of the quantized electric charge : @xmath77 where @xmath78 denotes the strength of the magnetic monopole .
the solution of this equation is easily found to be @xmath79 and the corresponding vector potential is given by @xmath80 where @xmath81 are the polar coordinates and @xmath82 is the unit vector in the @xmath83-direction .
the vector potential ( [ vectora ] ) reproduces the magnetic field ( [ bfield ] ) , except on the positive @xmath1-axis along which the magnetic field exhibits a singularity called the dirac string : @xmath84 where @xmath85 is the unit vector along the @xmath1-direction .
the dirac monopole can be created in the ferromagnetic phase of a spin-1 bec @xcite .
substituting @xmath86 , @xmath87 , and @xmath88 in eq .
( [ eqscb88 ] ) , we obtain @xmath89 \frac{e^{-i\varphi}}{\sqrt{2 } } \sin\theta \\[1 mm ] \sin^2 \frac{\theta}{2}. \end{pmatrix } \label{dirac}\end{aligned}\ ] ] the corresponding superfluid velocity @xmath90 [ defined by eq . ]
is the same as the vector potential ( [ vectora ] ) of the dirac monopole with the identification @xmath91 .
it is interesting to note that the order parameter ( [ dirac ] ) exhibits a doubly quantized vortex only along the positive @xmath1-axis .
one can say that because the first homotopy group of so(3 ) is @xmath32 , which allows only a singly quantized vortex called a polar - core vortex as a stable topological object , the doubly quantized vortex must terminate at the magnetic monopole .
in fact , the singular dirac monopole can deform continuously to a nonsingular spin texture if we take the following parametrization : @xmath92 \frac{e^{-i\varphi}}{\sqrt{2 } } \sin\beta \\[1 mm ] \sin^2 \frac{\beta}{2 } \end{pmatrix } , \ \
\ \beta(r,\varphi , z)=\theta ( 1-t)+\pi t. \label{dirac2}\end{aligned}\ ] ] we can see that as @xmath93 increases from 0 to 1 , the order parameter continuously changes from the dirac monopole at @xmath94 to a nonsingular texture @xmath95 at @xmath96 . when two different types of topological objects coexist , there exists an action of one object on the other as they move relative to each other . as an example , consider a spin-1 polar phase in which two monopoles with charge 1 merge in the presence of a half - quantum vortex .
as a monopole turns around the vortex , it changes sign .
thus , if they merge along line 1 , the order - parameter field must deform , as shown in fig . [
fig : homotopy9 ] ( b ) , and the combined object has the topological charge of two . on the other hand , when they merge along line 2 , the order - parameter manifold can deform into the one shown in fig .
[ fig : homotopy9 ] ( c ) , and the topological charge is zero , indicating that the two monopoles annihilate in pairs .
this example illustrates that the charge of the monopole changes its sign as it moves around the half - quantum vortex along curve 2 . *
shankar skyrmion * the third homotopy group characterizes topological objects called skyrmions that extend over the entire three - dimensional space .
a prime example of this is the so - called shankar monopole @xcite , although it is actually not a monopole but a skyrmion .
the topology of the shankar skyrmion is @xmath97 , which is supported by the ferromagnetic phase of a spin-1 bec .
the order parameter of the ferromagnetic bec is characterized by the direction of the local spin @xmath98 and the superfluid phase that can be parametrized in terms of two unit vectors such as @xmath99 and @xmath100 that form a triad with @xmath101 : @xmath102 .
concrete forms of @xmath98 and @xmath99 are introduced in sec .
[ sec : vortex_spin1ferro ] . the class one element of @xmath97 may be realized by rotating the triad at position @xmath57 about the direction @xcite @xmath103 through an angle @xmath104 , where @xmath105 and @xmath106 .
it follows from the last condition that the triad is uniform in spatial infinity , and thus , the three - dimensional space is compacted to @xmath107 .
the case of @xmath108 is shown in fig .
[ fig : homotopy8 ] .
the topological charge of the third homotopy group can be introduced in a manner similar to that of the second homotopy group .
now , the mapped vector has a four - component vector that is normalized to unity .
one such representation is @xmath109 the invariant of such a mapping is again given by the jacobian of the transformations : @xmath110 where @xmath111 and @xmath112 are the completely antisymmetric tensors of the third and fourth orders , respectively , and the roman letters run over @xmath113 and the greek letters run over @xmath114 . + * knot soliton * the third homotopy group also classifies knot solitons .
knots differ from other topological excitations such as vortices , monopoles , and shankar monopoles in that they are classified by a linking number while others are classified by winding numbers .
knots are characterized by mappings from a three - dimensional sphere @xmath107 to @xmath46 . as in the case of shankar monopoles ,
the @xmath107 domain is prepared by imposing a boundary condition that the order parameter takes on the same value in every direction at spatial infinity . here , we consider the spin-1 polar phase @xcite .
the order parameter manifold of the polar phase is given by @xmath115 , of which neither @xmath116 nor @xmath32 symmetry contributes to the homotopy groups in spaces higher than one dimension ; hence , we find that @xmath117 .
the associated integer topological charge @xmath118 is known as the hopf charge , and it is given by @xmath119 where @xmath120 @xcite .
note that the domain ( @xmath57 ) is three - dimensional , while the target space ( @xmath121 ) is two - dimensional .
consequently , the preimage of a point on the target @xmath46 constitutes a closed loop in @xmath107 , and the hopf charge is interpreted as the linking number of these loops : if the @xmath121 field has hopf charge @xmath118 , two loops corresponding to the preimages of any two distinct points on the target @xmath46 will be linked @xmath118 times [ see fig .
[ fig : knot_preimage ] ( a ) ] .
figure [ fig : knot_preimage ] ( b ) shows an example of the @xmath121 field of a polar bec with hopf charge 1 ( b ) is an _ unknot _ of a pair of rings with linking number 1 , because the preimage of one point on @xmath46 forms a simple unknotted ring . ] .
knots can be created by manipulating an external magnetic field . in the presence of an external magnetic field
, the linear zeeman effect causes the larmor precession of @xmath121 , whereas @xmath121 tends to become parallel to the magnetic field because of the quadratic zeeman effect .
suppose that we prepare an optically trapped bec in the @xmath122 state , i.e. , @xmath123 , by applying a uniform magnetic field in the @xmath1-direction .
then , we suddenly turn off the uniform field and switch on a quadrupole field . because of the linear zeeman effect
, @xmath121 starts rotating around the local magnetic field , and therefore , the @xmath121 field winds as a function of @xmath93 , resulting in the formation of knots .
figure [ fig : knottime ] shows the creation dynamics of knots in an optical trap subject to the quadrupole field , where the upper panels show the snapshots of the preimages of @xmath124 and @xmath125 and the bottom panels show cross sections of the density for @xmath126 ( bottom ) components on the @xmath127 plane .
the density pattern in @xmath126 components is characteristic of knots ; a double - ring pattern appears corresponding to one knot .
as the @xmath121 field winds in the dynamics , the number of rings increases .
this prediction can be tested by the stern - gerlach experiment .
topological objects are conjectured to play important roles in the formation of our universe . among them , knot solitons have recently attracted the interest of cosmologists since faddeev and niemi suggested that knots might exist as stable solitons in a three - dimensional classical field theory @xcite . |
this paper is concerned with the problem of estimating stationary ergodic processes with finite alphabet from a sample , an observed length @xmath0 realization of the process , with the @xmath1-distance being considered between the process and the estimated one .
the @xmath2-distance was introduced by ornstein @xcite and became one of the most widely used metrics over stationary processes .
two stationary processes are close in @xmath1-distance if there is a joint distribution whose marginals are the distributions of the processes such that the marginal processes are close with high probability ( see section [ secappl ] for the formal definition ) .
the class of ergodic processes is @xmath1-closed and entropy is @xmath1-continuous , which properties do not hold for the weak topology @xcite .
ornstein and weiss @xcite proved that for stationary processes isomorphic to i.i.d .
processes , the empirical distribution of the @xmath3-length blocks is a strongly consistent estimator of the @xmath3-length parts of the process in @xmath1-distance if and only if @xmath4 , where @xmath5 denotes the entropy of the process .
csiszr and talata @xcite estimated the @xmath0-length part of a stationary ergodic process @xmath6 by a markov process of order @xmath7 .
the transition probabilities of this markov estimator process are the empirical conditional probabilities , and the order @xmath8 does not depend on the sample .
they obtained a rate of convergence of the markov estimator to the process @xmath6 in @xmath1-distance , which consists of two terms .
the first one is the bias due to the error of the approximation of the process by a markov chain .
the second term is the variation due to the error of the estimation of the parameters of the markov chain from a sample . in this paper , the order @xmath7 of the markov estimator process is estimated from the sample .
for the order estimation , penalized maximum likelihood ( pml ) with general penalty term is used .
the resulted markov estimator process finds a tradeoff between the bias and the variation as it uses shorter memory for faster memory decays of the process @xmath6 .
if the process @xmath6 is a markov chain , the pml order estimation recovers its order asymptotically with a wide range of penalty terms
. not only an asymptotic rate of convergence result is obtained but also an explicit bound on the probability that the @xmath1-distance of the above markov estimator from the process @xmath6 is greater than @xmath9 .
it is assumed that the process @xmath6 is non - null , that is , the conditional probabilities of the symbols given the pasts are separated from zero , and that the continuity rate of the process @xmath6 is summable and the restricted continuity rate is uniformly convergent .
these conditions are usually assumed in this area @xcite .
the summability of the continuity rate implies that the process is isomorphic to an i.i.d .
process @xcite .
the above result on statistical estimation of stationary ergodic processes requires a non - asymptotic analysis of the markov order estimation for not necessarily finite memory processes . in this paper
, this problem is also investigated in more generality : under milder conditions than it would be needed for the above bound and not only for the pml method .
a popular approach to the markov order estimation is the minimum description length ( mdl ) principle @xcite .
this method evaluates an information criterion for each candidate order based on the sample and the estimator takes the order for which the value is minimal . the normalized maximum likelihood ( nml ) @xcite and the krichevsky
trofimov ( kt ) @xcite code lengths are natural information criteria because the former minimizes the worst case maximum redundancy for the model class of @xmath10-order markov chains , while the latter does so , up to an additive constant , with the average redundancy . the bayesian information criterion ( bic )
@xcite can be regarded as an approximation of the nml and kt code lengths .
the pml is a generalization of bic ; special settings of the penalty term yield the bic and other well - known information criteria , such as the akaike information criterion ( aic ) @xcite .
there are other methods for markov order estimation , see @xcite and references there , and the problem can also be formulated in the setting of hypothesis testing @xcite .
if a process is a markov chain , the nml and kt markov order estimators are strongly consistent if the candidate orders have an upper bound @xmath11 @xcite .
without such a bound , they fail to be consistent @xcite .
the bic markov order estimator is strongly consistent without any bound on the candidate orders @xcite .
if a process has infinite memory , the markov order estimators are expected to tend to infinity as @xmath12 .
the concept of context trees of arbitrary stationary ergodic processes is a model more complex than markov chains .
recent results @xcite in that area imply that this expectation holds true for the bic and kt markov order estimators but they provide no information about the asymptotics of the divergence . in this paper ,
the divergence of the pml , nml and kt markov order estimators for not necessarily finite memory processes is investigated .
not only asymptotic rates of divergence are obtained but also explicit bounds on the probability that the estimators are greater and less , respectively , than some order . instead of the usual assumption of non - nullness , it is assumed only that the conditional probabilities of one of the symbols given the pasts are separated from zero .
this property is called weakly non - nullness and is `` noticeably weaker '' than non - nullness @xcite .
first , the process is assumed to be weakly non - null and @xmath13-summable .
the @xmath14-summability @xcite is a condition weaker than the summability of the continuity rate . under these conditions ,
a bound on the probability that the estimators are greater than some order is obtained , that yields an @xmath15 upper bound on the estimated order eventually almost surely as @xmath12 . then , a bound on the probability that the estimators are less than some order is obtained assuming that the process is weakly non - null and the decay of its continuity rates is in some exponential range .
this bound implies that the estimators satisfying the conditions attain a @xmath16 divergence rate eventually almost surely as @xmath12 , where the coefficient @xmath17 depends on the range of the continuity rates . the class of processes with exponentially decaying continuity rate is considered in various problems @xcite
. fast divergence rate of the estimators are expected only for a certain range of continuity rates .
clearly , the estimators do not have a fast divergence rate if the memory decay of the process is too fast . on the other hand , too slow memory decay
is also not favored to a fast divergence rate because then the empirical probabilities do not necessarily converge to the true probabilities . to provide additional insight into the asymptotics of markov order estimators ,
the notion of consistent markov order estimation is generalized for infinite memory processes .
a markov order estimator is compared to its oracle version , which is calculated based on the true distribution of the process instead of the empirical distribution .
the oracle concept is used in various problems , see , for example , @xcite .
if the decay of the continuity rate of the process is faster than exponential , the ratio of the pml markov order estimator with sufficiently large penalty term to its oracle version is shown to converge to @xmath18 in probability .
the structure of the paper is the following . in section [ secnotation ] ,
notation and definitions are introduced for stationary ergodic processes with finite alphabets . in section [ secic ] , the pml , nml and kt information criteria are introduced .
section [ secmain ] contains the results on divergence of the information - criterion based markov order estimators . in section [ secappl ] ,
the problem of estimating stationary ergodic process in @xmath1-distance is formulated and our results are presented .
the results require bounds on empirical entropies , which are stated in section [ secmain ] and are proved in section [ secent ] .
section [ secproof ] contains the proof of the divergence results , and section [ secproofappl ] the proof of the process estimation results .
let @xmath19 be a stationary ergodic stochastic process with finite alphabet @xmath20 .
we write @xmath21 and @xmath22 for @xmath23 .
if @xmath24 , @xmath25 is the empty string . for two strings @xmath26 and @xmath27
, @xmath28 denotes their concatenation @xmath29 .
write @xmath30 and , if @xmath31 , @xmath32 for @xmath33 , @xmath34 .
the process @xmath6 is called _ weakly non - null _ if @xmath35 letting @xmath36 we say that the process @xmath6 is _
@xmath14-summable _ if @xmath37 the _ continuity rates _ of the process @xmath6 are @xmath38 and @xmath39 obviously , @xmath40 .
if @xmath41 , then the process @xmath6 is said to have _ summable continuity rate_. [ remgammaeq ] since for any @xmath42 and @xmath43 , @xmath44 , @xmath45 the above definition of continuity rate is equivalent to @xmath46 [ remgammaalpha ] the process is @xmath14-summable if it has summable continuity rate because @xmath47 the @xmath10-order _ entropy _ of the process @xmath6 is @xmath48 and the @xmath10-order _ conditional entropy _ is @xmath49 logarithms are to the base @xmath50 .
it is well known for stationary processes @xcite that the conditional entropy @xmath51 is a non - negative decreasing function of @xmath10 , therefore its limit exists as @xmath52 .
the _ entropy rate _ of the process is @xmath53 note that @xmath54 for any @xmath55 .
the process @xmath6 is a _ markov chain _ of order @xmath10 if for each @xmath56 and @xmath57 @xmath58 where @xmath59 is called initial distribution and @xmath60 is called transition probability matrix .
the case @xmath61 corresponds to i.i.d .
processes . the process @xmath6 is of _ infinite memory _ if it is not a markov chain for any order @xmath62 . for infinite memory processes , @xmath63 for any @xmath55 . in this paper
, we consider statistical estimates based on a sample @xmath64 , an @xmath0-length part of the process .
let @xmath65 denote the number of occurrences of the string @xmath66 in the sample @xmath64 @xmath67 for @xmath68 , the empirical probability of the string @xmath66 is @xmath69 and the empirical conditional probability of @xmath70 given @xmath66 is @xmath71 for @xmath61 , @xmath72 .
the @xmath10-order _ empirical entropy _ is @xmath73 and the @xmath10-order _ empirical conditional entropy _ is @xmath74 the likelihood of the sample @xmath64 with respect to a @xmath10-order markov chain model of the process @xmath6 with some transition probability matrix @xmath75 , by ( [ eqmcdef ] ) , is @xmath76 for @xmath77 , the _ maximum likelihood _ is the maximum in @xmath78 of the second factor above , which equals @xmath79 note that @xmath80 .
an information criterion assigns a score to each hypothetical model ( here , markov chain order ) based on a sample , and the estimator will be that model whose score is minimal .
[ defic ] for an information criterion @xmath81 the markov order estimator is @xmath82 here , the number of candidate markov chain orders based on a sample is finite , therefore the minimum is attained . if the minimizer is not unique , the smallest one will be taken as @xmath83 .
we consider three , the most frequently used information criteria , namely , the bayesian information criterion and its generalization , the family of penalized maximum likelihood ( pml ) , the normalized maximum likelihood ( nml ) code length @xcite , and the krichevsky trofimov ( kt ) code length @xcite . [ defpml ] given a penalty function @xmath84 , a non - decreasing function of the sample size @xmath0 , for a candidate order @xmath85 the pml criterion is @xmath86 the @xmath10-order markov chain model of the process @xmath6 is described by the conditional probabilities @xmath87 , and @xmath88 of these are free parameters . the second term of the pml criterion , which is proportional to the number of free parameters of the @xmath10-order markov chain model , is increasing in @xmath10 . the first term , for a given sample ,
is known to be decreasing in @xmath10 .
hence , minimizing the criterion yields a tradeoff between the goodness of fit of the sample to the model and the complexity of the model . if @xmath89 , the pml criterion is called _ bayesian information criterion _ ( bic ) , and if @xmath90 , _ akaike information criterion _ ( aic ) .
the minimum description length ( mdl ) principle minimizes the length of a code of the sample tailored to the model class . strictly speaking
, the information criterion would have an additive term , the length of a code of the structure parameter .
this additional term , the length of a code of @xmath10 , is omitted since it does not affect the results .
[ defnml ] for a candidate order @xmath85 , the nml criterion is @xmath91 where @xmath92 is the @xmath10-order nml - probability of @xmath64 .
writing @xmath93 the nml criterion can be regarded as a pml criterion in a broader sense .
[ defkt ] for a candidate order @xmath85 , the kt criterion is @xmath94 where @xmath95 } { ( n_{n-1}(a_1^k ) - 1 + { |a|}/{2 } ) ( n_{n-1}(a_1^k ) - 2 + { |a|}/{2 } ) \cdots ( { |a|}/{2 } ) } \ ] ] is the @xmath10-order kt - probability of @xmath64 .
( for @xmath61 , @xmath96 . )
the @xmath10-order kt - probability of the sample is equal to a mixture of the probabilities of the sample with respect to all @xmath10-order markov chains with uniform initial distribution , where the mixture distribution over the transition probability matrices @xmath97 is independent for the rows @xmath98 , @xmath99 , and has dirichlet @xmath100 distribution in the rows .
hence , the kt markov order estimator can be regarded as a bayes ( maximum a posteriori ) estimator .
the @xmath10-order nml and kt coding distributions are nearly optimal among the @xmath10-order markov chains , in the sense that the code lengths @xmath101 and @xmath102 minimize the worst case maximum and average , respectively , redundancy for this class ( up to an additive constant in the latter case ) .
the bic markov order estimator is strongly consistent @xcite , that is , if the process is a markov chain of order @xmath10 , then @xmath103 eventually almost surely as @xmath104 .
`` eventually almost surely '' means that with probability @xmath18 , there exists a threshold @xmath105 ( depending on the infinite realization @xmath106 ) such that the claim holds for all @xmath107 . increasing the penalty term , up to @xmath108 , where @xmath109 is a sufficiently small constant , does not affect the strong consistency .
it is not known whether or not the strong consistency holds for smaller penalty terms but it is known that if the candidate orders are upper bounded by @xmath110 , where @xmath109 is a sufficiently small constant , that is , the estimator minimizes the pml over the orders @xmath111 only , then @xmath112 still provides the strong consistency , where @xmath113 is a sufficiently large constant @xcite .
the nml and kt markov order estimators fail to be strongly consistent because for i.i.d .
processes with uniform distribution , they converge to infinity at a rate @xmath15 @xcite . however , if the candidate orders are upper bounded by @xmath11 , the strong consistency holds true @xcite .
if the process is of infinite memory , the bic and kt markov order estimators diverge to infinity @xcite . in this section ,
results on the divergence rate of the pml , nml and kt markov order estimators are presented .
bounds on the probability that the estimators are greater and less , respectively , than some order are obtained , with explicit constants .
the first implies that under mild conditions , the estimators do not exceed the @xmath15 rate eventually almost surely as @xmath12 .
the second bound implies that the rate @xmath114 is attained eventually almost surely as @xmath12 for the processes whose continuity rates decay in some exponential range . at the end of the section ,
the notion of consistent markov order estimation is generalized for infinite memory processes .
if the continuity rates decay faster than exponential , the pml markov order estimator is shown to be consistent with the oracle - type order estimate .
the proofs use bounds on the simultaneous convergence of empirical entropies of orders in an increasing set .
these bounds are obtained for finite sample sizes @xmath0 with explicit constants under mild conditions so they are of independent interest and are also presented here .
[ thentsmp ] for any weakly non - null and @xmath14-summable stationary ergodic process , for any @xmath115 @xmath116 and @xmath117 where @xmath118 are constants depending only on the distribution of the process . the proof including the explicit expression of the constants is in section [ secent ] .
the convergence of @xmath119 and @xmath120 , @xmath121 , to the entropy rate @xmath122 of the process could be investigated using theorem [ thentsmp ]
. however , good estimates of the entropy rate are known from the theory of universal codes . in particular , mixtures of the kt distributions over all possible orders provide universal codes in the class of all stationary ergodic processes @xcite , therefore the corresponding code length is a suitable estimate of the entropy rate . an application of the borel
cantelli lemma in theorem [ thentsmp ] yields the following asymptotic result . for any weakly non - null and @xmath14-summable stationary ergodic process , for any @xmath115 @xmath123 simultaneously for all @xmath124 , eventually almost surely as @xmath12 . by @xcite , under much stronger conditions on the process
, the convergence rate of @xmath125 and @xmath126 to @xmath122 is @xmath127 for some fixed @xmath128 .
hence , the rate in theorem [ thentsmp ] can not be improved significantly .
the first divergence result of the paper is the following .
[ thpmllarge ] for any weakly non - null and @xmath14-summable stationary ergodic process there exist @xmath129 depending only on the distribution of the process , such that for the markov order estimator @xmath130 @xmath131 for any sequence @xmath7 , @xmath132 , where ic is either the pml with arbitrary @xmath84 or the nml or the kt criterion . the proof including the explicit expression of the constants is in section [ secproof ] .
an application of the borel cantelli lemma in theorem [ thpmllarge ] yields the following asymptotic result .
[ copmllarge ] for any weakly non - null and @xmath14-summable stationary ergodic process there exists a constant @xmath113 such that for the markov order estimator @xmath130 @xmath133 eventually almost surely as @xmath12 , where ic is either the pml with arbitrary @xmath84 or the nml or the kt criterion .
the second divergence result is the following .
[ thpmllogsmp ] for any weakly non - null stationary ergodic process with continuity rates @xmath134 and @xmath135 for some @xmath136 ( @xmath137 ) , if @xmath138 the markov order estimator @xmath130 satisfies that @xmath139 if @xmath140 , where ic is either the pml with @xmath141 or the nml or the kt criterion , and @xmath142 , @xmath143 are constants depending only on the distribution of the process and @xmath84 . the proof including the explicit expression of the constants is in section [ secproof ] .
an application of the borel cantelli lemma in theorem [ thpmllogsmp ] yields the following asymptotic result .
[ corpmllog ] for any weakly non - null stationary ergodic process with continuity rates @xmath134 and @xmath135 for some @xmath136 with @xmath144 , the markov order estimator @xmath130 satisfies that @xmath145 eventually almost surely as @xmath12 , where ic is either the pml with @xmath146 or the nml or the kt criterion , and @xmath147 is a constant depending only on the distribution of the process .
the section concludes with the consistency result .
[ defopml ] for a candidate order @xmath85 the oracle pml criterion is @xmath148 for markov chains of order @xmath10 , @xmath149 if @xmath0 is sufficiently large , with any @xmath150 .
[ thoracle ] for any weakly non - null stationary ergodic process with @xmath151 the pml markov order estimator @xmath152 with @xmath153 , @xmath154 , is consistent in the sense that @xmath155 in probability as @xmath12 .
the proof is in section [ secproof ] .
in the results of this section , the divergence rate of markov order estimators will play a central role .
the problem of statistical estimation of stationary ergodic processes by finite memory processes is considered , and the following distance is used .
the per - letter hamming distance between two strings @xmath156 and @xmath157 is @xmath158 and the _ @xmath1-distance _ between two random sequences @xmath64 and @xmath159 is defined by @xmath160 where the minimum is taken over all the joint distributions @xmath161 of @xmath162 and @xmath163 whose marginals are equal to the distributions of @xmath64 and @xmath159 .
the process @xmath6 is estimated by a markov chain of order @xmath164 from the sample in the following way . the of a process @xmath6 based on the sample @xmath64 is the stationary markov chain , denoted by @xmath165 $ ] , of order @xmath10 with transition probability matrix @xmath166 . if the initial distribution of a stationary markov chain with these transition probabilities is not unique , then any of these initial distributions can be taken . in the previous section , weakly non - nullness is assumed for the process . in this section the process @xmath6 is assumed to be _ non - null _ , that is , @xmath167 [ rempmllarge ] for any non - null stationary ergodic process , @xmath168 for any @xmath99 .
hence , theorem [ thpmllarge ] holds with @xmath169 and @xmath170 , see the proof of the theorem .
the assumption of non - nullness allows us to use the following quantity instead of @xmath171 .
the _ restricted continuity rate _ of the process @xmath6 is @xmath172 similarly to remark [ remgammaeq ] , note that the above definition is equivalent to @xmath173 hence , @xmath174 for any fixed @xmath10 .
we say that the process @xmath6 has _ uniformly convergent restricted continuity rate _ with parameters @xmath175 , @xmath176 , @xmath177 if @xmath178 the order @xmath10 of the empirical markov estimator @xmath179 $ ] is estimated from the sample , using the pml criterion .
the estimated order needs to be bounded to guarantee an accurate assessment of the memory decay of the process .
[ deficr ] for an information criterion ic , the markov order estimator bounded by @xmath180 , @xmath181 , is @xmath182 the optimal order can be smaller than the upper bound if the memory decay of the process is sufficiently fast .
define @xmath183 where @xmath184 and @xmath185 . since @xmath186 is a decreasing function
, @xmath187 increases in @xmath0 but does not exceed @xmath188 .
it is less than @xmath188 if @xmath189 vanishes sufficiently fast , and then the faster @xmath189 vanishes , the slower @xmath187 increases .
the process estimation result of the paper is the following .
[ thqminsmp ] for any non - null stationary ergodic process with summable continuity rate and uniformly convergent restricted continuity rate with parameters @xmath175 , @xmath176 , @xmath177 , and for any @xmath190 , the empirical markov estimator of the process with the order estimated by the bounded pml markov order estimator @xmath191 , @xmath192 , with @xmath193 satisfies @xmath194_1^n \bigr ) > \frac{\beta_2}{{p_{\mathrm{inf}}}^{2 } } \max\biggl\ { \bar{\gamma } \biggl ( \biggl\lfloor\frac{\eta}{\theta_2 } \log n \biggr \rfloor\biggr ) , n^{- ( 1 - 4\eta\log({|a|^4}/{{p_{\mathrm{inf } } } } ) ) /({4\theta_1 } ) } \biggr\ } + \frac{1}{n^{1/2-\mu_n } } \biggr ) \\ & & \quad\le\exp\bigl ( -c_4 4^ { \mu_n\log n - |\log{p_{\mathrm{inf}}}| ( k_n ( \eta\log n , \bar{\gamma } , { c}\operatorname { pen}(n)/{n } ) + { \log\log n}/{\log|a| } ) } \bigr ) \\ & & \qquad { } + \exp\biggl ( -\frac{c_5 \eta^3}{\log n } n^{\eta2\log|a| } \biggr ) + 2^{-s_n \operatorname{pen}(n)},\end{aligned}\ ] ] if @xmath140 , where @xmath109 is an arbitrary constant , @xmath195 and @xmath196 are constants depending only on the distribution of the process . the proof including the explicit expression of the constants is in section [ secproofappl ] . if the process @xmath6 is a markov chain of order @xmath10 , then the restricted continuity rate is uniformly convergent with parameters @xmath197 , @xmath198 arbitrary ( arbitrarily close to @xmath18 ) , @xmath199 , and if @xmath0 is sufficiently large , @xmath200 and @xmath201 an application of the borel cantelli lemma in theorem [ thqminsmp ] yields the following asymptotic result .
[ cormain ] for any non - null stationary ergodic process with summable continuity rate and uniformly convergent restricted continuity rate with parameters @xmath175 , @xmath176 , @xmath177 , the empirical markov estimator of the process with the order estimated by the bounded pml markov order estimator @xmath202 with @xmath203 and @xmath204 satisfies @xmath205_1^n \bigr ) & \le&\frac{\beta_2}{{p_{\mathrm{inf}}}^{2 } } \max\biggl\ { \bar { \gamma } \biggl ( \biggl\lfloor\frac{r_n}{\theta_2 } \biggr\rfloor\biggr ) , n^{- { 1}/({4\theta_1 } ) } \biggr\}\\ & & { } + \frac { ( \log n)^{c_6 } } { \sqrt{n } } 2^ { \bar{\gamma } , { c}\operatorname{pen}(n)/{n } ) } \end{aligned}\ ] ] eventually almost surely as @xmath12 , where @xmath109 is an arbitrary constant , and @xmath206 are constants depending only on the distribution of the process . if the memory decay of the process is slow , the first term in the bound in corollary [ cormain ] , the bias , is essentially @xmath207 , and the second term , the variance , is maximal . if the memory decay is sufficiently fast , then the rate of the estimated order @xmath208 and the rate of @xmath187 are smaller , therefore the variance term is smaller , while the bias term is smaller as well .
the result , however , shows the optimality of the pml markov order estimator in the sense that it selects an order which is small enough to allow the variance to decrease but large enough to keep the bias below a polynomial threshold .
in this section , we consider the problem of simultaneous convergence of empirical entropies of orders in an increasing set , and prove the following theorem that formulates theorem [ thentsmp ] with explicit constants .
[ thent ] for any weakly non - null and @xmath14-summable stationary ergodic process , for any @xmath115 @xmath209 and @xmath210 first , we show the following bounds .
[ prent ] for any weakly non - null and @xmath14-summable stationary ergodic process , for any @xmath211 and @xmath212 , @xmath213 and @xmath214 fix @xmath215 . applying lemma [ lemsch ] in the to the distributions @xmath216 and @xmath217 , @xmath218 d_{{\mathrm{tv } } } ( \hat{p}_k , p_k ) , \ ] ] if @xmath219 . for any @xmath220 ,
the right of ( [ eqsch ] ) can be written as @xmath221 \\ & & \quad\le\frac{k \log|a|}{\log { \mathrm{e } } } d_{{\mathrm{tv } } } ( \hat{p}_k , p_k ) + \frac{1}{{\mathrm{e } } } \frac{1+\nu}{\nu } d_{{\mathrm{tv}}}^{{1}/({1+\nu } ) } ( \hat{p}_k , p_k ) , \nonumber\end{aligned}\ ] ] where we used the bound @xmath222 , @xmath223 . by @xcite , for any string @xmath99 and @xmath224 , @xmath225 where @xmath226 is positive for any weakly non - null and @xmath14-summable stationary ergodic process .
( [ eqgl ] ) implies that @xmath227\\[-8pt ] & \le & { \mathrm{e}}^{1/{\mathrm{e } } } |a|^k \exp\biggl ( \frac{-c_{\alpha}(n - k+1 ) t^2 } { k |a|^{2k } } \biggr).\nonumber\end{aligned}\ ] ] applying ( [ eqgl2 ] ) to ( [ eqsch2 ] ) , @xmath228 this completes the proof of the first claimed bound as @xmath229 the second claimed bound follows using @xmath230 and @xmath231 as @xmath232 now , the theorem follows from the proposition with special settings .
proof of theorem [ thent ] we use proposition [ prent ] setting @xmath233 , @xmath234 , and @xmath235 . then , in the exponent of the first inequality of the proposition , @xmath236 where we used that @xmath115 .
this gives the lower bound @xmath237 on the exponent and completes the proof of the first claimed bound .
the second claimed bound follows similarly from the second inequality of the proposition with the same settings .
in this section , we consider the divergence of the pml , nml and kt markov order estimators and prove theorems [ thpmllarge ] , [ thpmllogsmp ] and [ thoracle ] .
proof of theorem [ thpmllarge ] by @xcite , any weakly non - null and @xmath14-summable process is @xmath238-mixing with a coefficient related to @xmath239 and @xmath240 .
namely , there exists a sequence @xmath241 , @xmath242 , satisfying @xmath243 such that for each @xmath10 , @xmath244 , @xmath245 and each @xmath99 , @xmath246 , with @xmath247 , @xmath248 this implies that for any @xmath249 @xmath250 since @xmath251 and @xmath252 , @xmath253 for sufficiently large @xmath254 .
then @xmath255 holds with @xmath256 and @xmath257 .
thus , for any @xmath10 , @xmath258 for any information criterion ic , we can write @xmath259\\[-8pt ] & & \quad\subseteq\bigl\ { \mathrm{ic}_{x_1^n } ( m )
< \mathrm { ic}_{x_1^n } ( k_n ) \mbox { for some } m > k_n \bigr\ } \cap\bigl\ { n_n\bigl(a_1^{k_n}\bigr ) \le1 \mbox { for all } a_1^{k_n } \bigr\ } \nonumber \\ & & \qquad { } \cup\bigl\ { n_n\bigl(a_1^{k_n } \bigr ) \ge2 \mbox { for some } a_1^{k_n } \bigr\}.\nonumber\end{aligned}\ ] ] here , @xmath260 for all @xmath261 implies that @xmath262 for all @xmath263 for all @xmath264 , which further implies that for all @xmath265 ( i ) @xmath266 and therefore @xmath267 and @xmath268 and ( ii ) @xmath269 .
then all the three information criteria do not depend on the sample and are non - decreasing in @xmath244 .
hence , in ( [ eqsplit ] ) @xmath270 is an empty set .
thus , ( [ eqsplit ] ) gives @xmath271 and using ( [ eqexp ] ) completes the proof . to prove theorem [ thpmllogsmp ]
, first we show the following bounds .
[ thpmlh ] for any weakly non - null and @xmath14-summable stationary ergodic process with @xmath272 for some @xmath273 , if @xmath274 the pml markov order estimator @xmath152 satisfies that @xmath275 if @xmath276 , where @xmath277 the markov order estimator @xmath130 , where ic is either nml or kt , satisfies that @xmath278 if @xmath279 , where @xmath280 for markov chains of order @xmath10 , in proposition [ thpmlh ] @xmath281 if @xmath0 is sufficiently large .
proof of proposition [ thpmlh ] let @xmath115 be arbitrary and @xmath282 for any information criterion ic , we can write for any @xmath283 @xmath284\\[-8pt ] & & \quad\subseteq\biggl ( \biggl\ { \mathrm{ic}_{x_1^n } ( m ) \le \mathrm{ic}_{x_1^n } \biggl ( \biggl\lfloor { \frac { \varepsilon\log n}{4\log|a| } } \biggr \rfloor\biggr ) \mbox { for some } m < k_n \biggr\ } \cap b_n \biggl ( { \frac{\varepsilon\log n}{4\log|a| } } \biggr ) \biggr)\nonumber\\ & & \qquad { } \cup \overline{b_n \biggl ( { \frac { \varepsilon \log n}{4\log|a| } } \biggr)}.\nonumber\end{aligned}\ ] ] \(i ) if @xmath285 , by the definition of the pml information criterion , see definition [ defpml ] , @xmath286 & & \hspace*{17pt}\quad\le\bigl(|a|-1\bigr ) \bigl ( |a|^ { \lfloor { ( { \varepsilon \log n})/({4\log|a| } ) } \rfloor } - |a|^m \bigr ) \operatorname{pen}(n ) \mbox { for some } m < k_n \biggr\ } \cap b_n \biggl ( { \frac{\varepsilon\log n}{4\log|a| } } \biggr ) \nonumber \\[-2pt ] & & \quad\subseteq\biggl\ { \hat{h}_{m}\bigl(x_1^n \bigr ) - \hat{h } _ { \lfloor ( { \varepsilon\log n})/({4\log|a| } ) \rfloor}\bigl(x_1^n\bigr ) \nonumber\\[-9pt]\\[-9pt ] & & \hspace*{25.5pt}\le\bigl(|a|-1\bigr ) |a|^ { \lfloor { ( { \varepsilon\log n})/({4\log|a| } ) } \rfloor } \frac { \operatorname{pen}(n)}{n- \lfloor(\varepsilon\log n)/(4\log|a| ) \rfloor } \mbox { for some } m < k_n \biggr\ } \nonumber\\[-2pt ] & & \qquad{}\cap b_n \biggl ( { \frac{\varepsilon\log n}{4\log|a| } } \biggr ) \nonumber \\[-2pt ] & & \quad\subseteq\biggl\ { h_m - h _ { \lfloor({\varepsilon\log n})/({4\log \frac{(|a|-1 ) |a|^{({\varepsilon\log n})/({4\log|a| } ) } \operatorname{pen}(n)}{n- ( \varepsilon\log n)/(4\log & & \hspace*{23.5pt}\mbox { for some } m <
k_n \biggr\}.\nonumber\end{aligned}\ ] ] since for any @xmath115 @xmath287 we have @xmath288 now , let @xmath289 and @xmath7 be as in the claim of the proposition . using the conditions @xmath272 and @xmath290 , @xmath291 thus ,
if @xmath276 , it follows that @xmath292 , and for any @xmath293 @xmath294\\[-9pt ] & \ge&(h_{k_n-1 } - \bar{h } ) - \frac{1}{\sqrt{n } } \ge\frac { 3\max(\sqrt { n},(|a|-1 ) \operatorname{pen}(n ) ) } { n^{1-\varepsilon } } , \quad \nonumber\end{aligned}\ ] ] where we used that @xmath51 is non - increasing . comparing ( [ eqhpmllb ] ) to ( [ eqhpmlub ] ) , the right of ( [ eqsplitpml ] ) is an empty set , and ( [ eqsplit2 ] ) yields @xmath295 if @xmath276 , according to theorem [ thent ] .
\(ii ) if @xmath296 , by the definition of the nml information criterion , see definition [ defnml ] , @xmath297\\[-8pt ] & & \qquad\hspace*{4pt}\mbox { for some } m < k_n \biggr\ } \nonumber \\ & & \qquad{}\cap b_n \biggl ( { \frac{\varepsilon\log n}{4\log|a| } } \biggr ) \nonumber \\ & & \quad\subseteq\biggl\ { h_m - h _ { \lfloor({\varepsilon\log n})/({4\log \frac { \log\sigma ( n , { \lfloor({\varepsilon\log n})/({4\log|a| } ) \rfloor } ) } { n- ( \varepsilon\log n)/(4\log|a| ) } + \frac{2}{n^ { 1/2-\varepsilon } } \nonumber\\ & & \qquad\hspace*{4pt}\mbox { for some } m < k_n \biggr \},\nonumber\end{aligned}\ ] ] where in the second relation we used that @xmath298 for any @xmath299 . by lemma [ lemktml ] in the , @xmath300 that gives the upper bound @xmath301 using ( [ eqsigmaub ] ) and ( [ eqatom ] ) , @xmath302 using @xmath303 , @xmath223
, it follows that @xmath304 if @xmath305 , which implies that @xmath306 thus , the expression in ( [ eqsplitnml ] ) can be bounded as @xmath307\\[-8pt ] & & \quad < \frac { 3}{n^{1/2-\varepsilon } } \qquad\mbox{if } n\ge\max\bigl\ { 24\bigl(\log^4 { \mathrm{e}}\bigr ) \bigl(|a|-1\bigr)^4 , 4c_{\mathrm { kt}}^2 \bigr\}.\nonumber\end{aligned}\ ] ] now , let @xmath289 and @xmath7 be as in the claim of the proposition .
then the conditions @xmath272 and @xmath290 imply ( [ eqdhpmlub ] ) , thus , if @xmath276 , it follows that @xmath308 , and for any @xmath293 @xmath309 where we used that @xmath51 is non - increasing . comparing ( [ eqhnmllb ] ) to ( [ eqhnmlub ] )
, the right of ( [ eqsplitnml ] ) is an empty set , and ( [ eqsplit2 ] ) yields @xmath310 if @xmath311 , according to theorem [ thent ] .
\(iii ) if @xmath312 , by the definition of the kt information criterion , see definition [ defkt ] , and using that @xmath313 for any @xmath314 , @xmath315 & & \quad\subseteq\biggl\ { ( n - m)\hat{h}_{m}\bigl(x_1^n \bigr ) - \biggl ( n- \biggl\lfloor { \frac{\varepsilon\log n}{4\log|a| } } \biggr\rfloor\biggr ) \hat { h } _ { \lfloor({\varepsilon\log n})/({4\log|a| } ) \rfloor}\bigl(x_1^n\bigr ) \nonumber \\[-0.8pt ] & & \hspace*{16.4pt}\quad\le\log\mathrm{ml } _ { \lfloor({\varepsilon\log n})/({4\log \bigr ) - \log p_{\mathrm{kt } , \lfloor ( { \varepsilon\log n})/({4\log|a| } ) \rfloor } \bigl(x_1^n\bigr ) \mbox { for some } m < k_n \biggr\ } \nonumber\\[-0.8pt ] & & \qquad{}\cap b_n \biggl ( { \frac{\varepsilon\log n}{4\log|a| } } \biggr ) \nonumber \\[-0.8pt ] & & \quad\subseteq\biggl\{\hat{h}_{m}\bigl(x_1^n \bigr ) - \hat{h } _ { \lfloor({\varepsilon\log n})/({4\log|a| } ) \rfloor}\bigl(x_1^n\bigr ) \nonumber\\[-0.8pt ] & & \hspace*{15.6pt}\quad\le\frac { \log\mathrm{ml } _ { \lfloor({\varepsilon\log n})/({4\log|a| } ) \rfloor}(x_1^n ) - \log p_{\mathrm{kt } , \lfloor({\varepsilon\log n})/({4\log|a| } ) \rfloor } ( x_1^n ) } { n- \lfloor(\varepsilon\log n)/(4\log|a| ) \rfloor } \\[-0.8pt ] & & \qquad\hspace*{4.5pt}\mbox { for some } m < k_n \biggr\ } \nonumber\\[-0.8pt ] & & \qquad{}\cap b_n \biggl ( { \frac{\varepsilon\log n}{4\log|a| } } \biggr ) \nonumber \\[-0.8pt ] & & \quad\subseteq\biggl\ { h_m - h _ { \lfloor({\varepsilon\log n})/({4\log & & \hspace*{15.6pt}\quad\le \frac { \log\mathrm{ml } _ { \lfloor ( { \varepsilon\log n})/({4\log|a| } )
\rfloor}(x_1^n ) - \log p_{\mathrm{kt } , \lfloor({\varepsilon\log n})/({4\log|a| } ) \rfloor } ( x_1^n ) } { n- ( \varepsilon\log n)/(4\log|a| ) } + \frac{2}{n^{1/2-\varepsilon } } \nonumber \\[-0.8pt ] & & \qquad\hspace*{4.5pt } \mbox { for some } m < k_n \biggr\}.\nonumber\end{aligned}\ ] ] by lemma [ lemktml ] in the , @xmath316 & & \quad\le c_{\mathrm{kt } } |a|^{({\varepsilon\log n})/({4\log|a| } ) } + \frac { |a|-1}{2 } { n}{|a|^ { \lfloor({\varepsilon\log n})/({4\log|a| } ) \rfloor}},\end{aligned}\ ] ] and the proof continues in the same way as in the nml case ( ii ) .
now , we are ready to prove theorem [ thpmllogsmp ] .
we prove the following theorem that formulates theorem [ thpmllogsmp ] with explicit constants .
[ thpmllog ] for any weakly non - null stationary ergodic process with continuity rates @xmath134 and @xmath135 for some @xmath136 ( @xmath137 ) , if @xmath317 the pml markov order estimator @xmath152 satisfies that @xmath318 if @xmath319 , where @xmath320 the markov order estimator @xmath130 , where ic is either nml or kt , satisfies that @xmath321 if @xmath322 , where @xmath323 [ here , @xmath324 is the constant in the well - known bound of @xmath325 , see lemma in the . ] by remark [ remgammaalpha ] , @xmath326 implies the @xmath14-summability .
the deviation of the conditional entropies from the entropy rate will also be controlled by the continuity rates of the process , and proposition [ thpmlh ] will yield the claim of the theorem .
first , for any @xmath327 , @xmath328 on the right of ( [ eqgsplit1 ] ) , the difference of entropies of the conditional distributions @xmath329 and @xmath330 appears . by remark [ remgammaeq ] ,
the total variation of these conditional distributions can be upper bounded as @xmath331 hence , applying lemma [ lemsch ] in the it follows , similar to the bound ( [ eqsch ] ) and ( [ eqsch2 ] ) in the proof of proposition [ prent ] , that @xmath332\\[-8pt ] & & \quad\le\frac{\log|a|}{\log { \mathrm{e } } } \bar{\gamma}(k ) + \frac{1}{{\mathrm{e } } } \frac{1+\nu}{\nu } \bar{\gamma}(k)^ { { 1}/({1+\nu } ) } \nonumber \\ & & \quad\le\frac{2\log|a|}{\log { \mathrm{e } } } \frac{1+\nu}{\nu } \bar{\gamma } ( k)^{{1}/({1+\nu})}\nonumber\end{aligned}\ ] ] for any @xmath220 , if @xmath333 . setting @xmath334 , combining ( [ eqgsplit11 ] ) with ( [ eqgsplit1 ] ) and taking @xmath335 yield the bound @xmath336 if @xmath333 .
since @xmath337 , the bound ( [ eqklb ] ) is trivial if @xmath338 .
hence , using the assumption @xmath134 of the theorem , @xmath339 and the assumption @xmath272 of proposition [ thpmlh ] is satisfied with @xmath340 thus , the constraint @xmath290 in proposition [ thpmlh ] becomes @xmath341 , and @xmath342 becomes @xmath343 next , for any @xmath344 , @xmath345 where @xmath346 denotes the kullback
leibler divergence . using pinsker s inequality @xcite , ( [ eqgsplit2 ] )
can be lower bounded by @xmath347 where in the last inequality we used the assumption @xmath348 of the theorem .
hence , in case ( i ) @xmath349 while in case ( ii ) @xmath350 and the proof is completed . finally , we prove the following proposition that directly implies theorem [ thoracle ] .
[ proporacle ] for any weakly non - null stationary ergodic process with continuity rate @xmath351 , @xmath352 , and for any @xmath353 , if @xmath354 is so small and @xmath355 is so large that @xmath356 and @xmath357 the pml markov order estimator @xmath152 with @xmath153 satisfies that @xmath358 if @xmath0 is sufficiently large , where @xmath359 is a constant depending only on the distribution of the process .
the proof of theorem [ thpmllog ] begins with the observation that the summability of the continuity rate implies the @xmath13-summability .
hence , the conditions of theorem [ thent ] are satisfied now . moreover , according to ( [ eqhg ] ) , @xmath360 also implies that @xmath361 set @xmath362 , @xmath289 and @xmath363 as in the conditions of the proposition , and define a sequence @xmath364 such that for sufficiently large @xmath0@xmath365 due to ( [ eqhg2 ] ) , such a sequence exists . since @xmath366 is non - negative decreasing , it is sufficient to show this when @xmath367 . then , writing @xmath7 in the form @xmath368 , @xmath369 if @xmath0 is sufficiently large , that implies ( i ) and ( ii ) if @xmath370 such @xmath220 exists because it follows from the condition @xmath371 that @xmath372 . moreover , the condition @xmath373 in ( [ eqbndef ] ) in the proof of proposition [ thpmlh ] . similar to ( [ eqsplit2 ] ) and ( [ eqsplitpml ] ) , we can write that @xmath374 that is empty set by ( i ) , if @xmath0 is large enough and @xmath292 .
the latter is satisfied because of . on the other hand , @xmath375 & & \quad\subseteq\biggl\ { \mathrm{pml}_{x_1^n } ( m )
< \mathrm { pml}_{x_1^n } ( k_n ) \mbox { for some } ( 1+\xi/2 ) k_n < m\le\frac{\varepsilon\log n}{4\log|a| } \biggr\}\nonumber\\[-0.5pt ] & & \qquad { } \cap b_n \biggl ( { \frac{\varepsilon\log n}{4\log|a| } } \biggr ) \nonumber \\[-0.5pt ] \label{eqoupperhalf } & & \quad\subseteq\biggl\ { \mathrm{pml}_{o , n } ( m ) - \frac { 2n}{n^{1/2-\varepsilon } } < \mathrm{pml}_{o , n } ( k_n ) \mbox { for some } m>(1+\xi/2 ) k_n \biggr\ } \\[-0.5pt ] & & \quad\subseteq\biggl\ { \bigl(|a|-1\bigr ) \bigl ( |a|^m - |a|^{k_n } \bigr ) \operatorname{pen}(n ) - \frac{2n}{n^{1/2-\varepsilon } } < ( n- k_n ) h_{k_n } - ( n - m ) h_{m } \nonumber \\[-0.5pt ] & & \hspace*{25pt } \mbox { for some } m>(1+\xi/2 ) k_n \biggr\ } \nonumber \\[-0.5pt ] & & \quad\subseteq\biggl\ { \bigl(|a|-1\bigr ) \bigl ( |a|^m - |a|^{k_n } \bigr ) \frac { \operatorname{pen}(n)}{n } - \frac{2}{n^{1/2-\varepsilon } } < h_{k_n } - \biggl(1- \frac{m}{n } \biggr ) h_{m } \nonumber \\[-0.5pt ] & & \hspace*{25pt } \mbox { for some } m>(1+\xi/2 ) k_n \biggr\ } \nonumber \\[-0.5pt ] & & \quad\subseteq\biggl\ { \bigl(|a|-1\bigr ) \bigl ( |a|^m - |a|^{k_n } \bigr ) \frac { \operatorname{pen}(n)}{n } - \frac{2}{n^{1/2-\varepsilon } } - \frac { m}{n}\bar{h } \nonumber\\[-0.5pt ] & & \hspace*{26.5pt } < ( h_{k_n } - \bar{h } ) - \biggl(1-\frac{m}{n } \biggr ) ( h_{k_n } - \bar{h } ) \nonumber \\[-0.5pt ] & & \hspace*{25pt } \mbox { for some } m>(1+\xi/2 ) k_n \biggr\ } \nonumber \\[-0.5pt ] \label{eqoupper1 } & & \quad\subseteq\biggl\ { h_{k_n } - \bar{h } > \frac { ( |a|-1)^2 } { 2 } n^{-1+\kappa } \biggr\}\end{aligned}\ ] ] that is empty set by ( ii ) , if @xmath0 is large enough
. observe that @xmath376 if @xmath0 is sufficiently large .
indeed , on indirect way the following sequence of implications can be written @xmath377 that does not hold by ( [ eqolowerhalf ] ) and ( [ eqolower1 ] ) if @xmath0 is large enough , and @xmath378 that does not hold either by ( [ eqoupperhalf ] ) and ( [ eqoupper1 ] ) if @xmath0 is large enough . finally , using ( [ eqknko ] ) , we get @xmath379 where the first two terms are zero if @xmath0 is large enough by ( [ eqolowerb])([eqolower1 ] ) and ( [ eqoupperb])([eqoupper1 ] ) . using proposition [ propqminupper ] with @xmath380 , @xmath381 and @xmath382 , @xmath383 because @xmath384 but @xmath385 according to the condition @xmath386
. then the claim of the proposition follows from theorem [ thent ] .
in this section , we consider the estimation of stationary ergodic processes by finite memory processes .
first , define @xmath387 and @xmath388 clearly , if @xmath389 , then @xmath390 .
now we prove the following theorem that formulates theorem [ thqminsmp ] with explicit constants .
[ thqmin ] for any non - null stationary ergodic process with summable continuity rate and uniformly convergent restricted continuity rate with parameters @xmath175 , @xmath176 , @xmath177 , for any @xmath190 , the empirical markov estimator of the process with the order estimated by the bounded pml markov order estimator @xmath391 , @xmath192 , with penalty function @xmath392 satisfies @xmath393_1^n \bigr ) > \frac{\beta_2}{{p_{\mathrm{inf}}}^{2 } } g_n + \frac{1}{n^{1/2-\mu_n } } \biggr ) \\ & & \quad\le2 { \mathrm{e}}^{1/{\mathrm{e } } } |a|^{k_n+h_n+2 } \exp\biggl\ { - \frac { { p_{\mathrm{inf}}}^2 } { 16 { \mathrm{e}}|a|^3 ( \alpha+ { p_{\mathrm{inf } } } ) ( \beta_1 + 1)^2 } \frac { ( n - k_n - h_n ) } { ( 1+k_n+h_n ) n } \\ & & \hspace*{87.4pt}\qquad{}\times 4^{-(k_n+h_n ) |\log{p_{\mathrm{inf}}}| } \biggl [ 4^{\mu_n\log n } - \frac { ( k_n+h_n)|\log{p_{\mathrm{inf}}}|(\beta_1 + 1)^2}{2 } \biggr ] \biggr\ } \\ & & \qquad{}+ 12 { \mathrm{e}}^{1/{\mathrm{e } } } \exp\biggl ( -\frac{7\alpha_0 ( \log|a|)^3 \eta^3}{4{\mathrm{e}}(\alpha+\alpha_0 ) } \frac { n^{\eta2\log|a| } } { \log n } + \bigl(\eta\log|a|\bigr)\log n \biggr ) \\ & & \qquad{}+ \exp\biggl ( -\bigl(|a|-1\bigr ) |a|^{k_n+h_n+1 } \\ & & \hspace*{32pt}\qquad{}\times\operatorname{pen}(n ) \biggl [ 1 - \frac { 1}{|a|^{1+h_n } } - \frac{1}{2\operatorname{pen}(n ) } \bigl ( \log n - ( k_n+h_n ) \log|a| \bigr ) \biggr ] \\ & & \hspace*{32pt}\qquad { } + \frac{c \operatorname{pen}(n)}{{p_{\mathrm{inf}}}/\log { \mathrm{e } } } + \eta\log n ) \biggr),\end{aligned}\ ] ] if @xmath0 is so large that @xmath394 where @xmath395 and @xmath109 is an arbitrary constant and @xmath396 is an arbitrary sequence . the proof is based on the following two propositions .
[ propqmin ] for any non - null and @xmath14-summable stationary ergodic process with uniformly convergent restricted continuity rate with parameters @xmath397 , @xmath176 , @xmath177 , the bounded pml markov order estimator @xmath398 with penalty function @xmath141 satisfies that @xmath399 if @xmath0 is so large that @xmath400 , where @xmath401 the bonded markov order estimator @xmath402 if @xmath0 is so large that @xmath400 and @xmath403 , where @xmath404 first , define @xmath405 similar to ( [ eqbndef ] ) in the proof of proposition [ thpmlh ] .
similar to ( [ eqsplit2])([eqhpmlub ] ) , we can write for any @xmath406 that @xmath407 now , the difference @xmath408 in ( [ eqsplit22 ] ) is controlled as follows . for any @xmath409 , @xmath410\\[-8pt ] & & \quad=\sum_{a_1^k \in a^k } p\bigl(a_1^k \bigr ) \sum_{a\in a } p\bigl(a| a_1^k \bigr ) \log\frac{p(a| a_1^k ) } { p(a| a_{k - m+1}^k ) } \nonumber \\ & & \quad=\sum_{a_1^k \in a^k } p\bigl(a_1^k \bigr ) d \bigl ( p\bigl ( \cdot| a_1^k \bigr ) \| p\bigl ( \cdot| a_{k - m+1}^k \bigr ) \bigr).\nonumber\end{aligned}\ ] ] using pinsker s inequality @xcite , ( [ eqgsplit22a ] ) can be lower bounded by @xmath411 using ( [ eqgsplit222 ] ) and the assumption @xmath412 if @xmath413 ( @xmath414 ) , it follows that @xmath415 hence , we can write @xmath416 let @xmath7 be as in the claim of the proposition and suppose that @xmath417 .
then , since @xmath51 is non - increasing , for any @xmath418 @xmath419 ) to ( [ eqsplit22 ] ) , the first term on the right in ( [ eqsplit22 ] ) equals zero , therefore @xmath420 by theorem [ thent ] with @xmath421 . in cases @xmath422 and @xmath423 , the proofs deviate from the above similar to as ( ii ) and ( iii ) deviate from ( i ) in the proof of proposition [ thpmlh ] .
now , instead of ( [ eqhnmlub ] ) we have @xmath424 [ propqminupper ] for any non - null stationary ergodic process , the bounded pml markov order estimator @xmath425 satisfies that @xmath426 \biggr)\end{aligned}\ ] ] for any @xmath427 . for any @xmath428 ,
@xmath429 using @xmath430 and @xmath431 , ( [ eqpfact1 ] ) can be upper bounded by @xmath432 now , let @xmath433 . by the definition of the pml information criterion , see definition [ defpml ] , for any @xmath434 @xmath435\\[-8pt ] & & { } + \bigl(|a|-1\bigr)|a|^{m_n } \operatorname{pen}(n ) \qquad\mbox{if } x_1^n \in c_{n , k } .
\nonumber\end{aligned}\ ] ] by lemma [ lemktml ] in the , @xmath436 combining ( [ eqpfact2 ] ) , ( [ eqsh1 ] ) and ( [ eqktml ] ) , @xmath437 that implies @xmath438 \biggr),\nonumber\end{aligned}\ ] ] where in the last inequality we used @xmath439 , @xmath223 . in the exponent of ( [ eqsh2 ] ) , it may be assumed that @xmath440 is multiplied by a negative number otherwise the bound is trivial .
then , the claim of the lemma follows from ( [ eqsh2 ] ) as @xmath441 now , we are ready to prove theorem [ thqmin ] .
proof of theorem [ thqmin ] letting @xmath442 and @xmath443_1^n \bigr ) > \frac{\beta_2}{{p_{\mathrm{inf}}}^{2 } } g_n + \frac{1}{n^{1/2-\mu_n } } \biggr ) \nonumber \\ & & \quad\le{\operatorname{pr}}\biggl ( \biggl\ { \bar{d } \bigl ( x_1^n , \hat{x } \bigl [ \hat{k}_{\mathrm{pml } } \bigl(x_1^n | \eta\log n\bigr ) \bigr]_1^n \bigr ) > \frac{\beta_2}{{p_{\mathrm{inf}}}^{2 } } g_n + \frac{1}{n^{1/2-\mu_n } } \biggr\ } \cap g_n
\cap h_n \biggr ) \nonumber \\ & & \qquad { } + { \operatorname{pr}}(\bar{g}_n ) + { \operatorname{pr}}(\bar { h}_n ) \\ & & \quad\le{\operatorname{pr}}\biggl ( \biggl\ { \bar{d } \bigl ( x_1^n , \hat{x } \bigl [ \hat{k}_{\mathrm{pml } } \bigl(x_1^n | \eta\log n\bigr ) \bigr]_1^n \bigr ) > \frac { \beta_2}{{p_{\mathrm{inf}}}^{2 } } \bar{\gamma } \bigl ( \hat{k}_{\mathrm{pml } } \bigl(x_1^n \biggr ) \nonumber \\ & & \qquad { } + { \operatorname{pr}}(\bar{g}_n ) + { \operatorname{pr}}(\bar { h}_n ) .\nonumber\end{aligned}\ ] ] the three terms on the right of ( [ eqfinal ] ) is bounded as follows .
since the process is non - null with summable continuity rate , lemma [ lemapprox ] in the with @xmath444 , @xmath445 and @xmath446 gives @xmath447_1^n \bigr ) > \frac { \beta_2}{{p_{\mathrm{inf}}}^{2 } } \bar{\gamma } \bigl ( \hat{k}_{\mathrm{pml } } \bigl(x_1^n \biggr ) \nonumber \\ & & \quad\le2 { \mathrm{e}}^{1/{\mathrm{e } } } |a|^{k_n+2 } \exp\biggl\ { - \frac { { p_{\mathrm{inf}}}^2 } { 16 { \mathrm{e}}|a|^3 ( \alpha+ { p_{\mathrm{inf } } } ) ( \beta_1 + 1)^2 } \frac { ( n - k_n ) 4^{-k_n |\log{p_{\mathrm{inf}}}| } } { ( 1+k_n ) n } \\ & & \hspace*{90pt}{}\times \biggl [ 4^{\mu_n\log n } - \frac{k_n|\log{p_{\mathrm{inf}}}|(\beta_1 + 1)^2}{2 } \biggr ] \biggr\ } .\nonumber\end{aligned}\ ] ] by remark [ remgammaalpha ] , the summability of the continuity rate implies the @xmath14-summability .
hence , for the non - null process with summable continuity rate and uniformly convergent restricted continuity rate with parameters @xmath175 , @xmath176 , @xmath177 , proposition [ propqmin ] implies that @xmath448 if ( [ eqnbound ] ) holds because @xmath449 applying proposition [ propqminupper ] with @xmath450 , @xmath451 and @xmath452 , it follows that @xmath453 \\ & & \hspace*{63pt}{}+ \frac{c \operatorname{pen}(n)}{{p_{\mathrm{inf}}}/\log { \mathrm{e } } } + |a|^{k_n+1 } c_{\mathrm { kt } } + \log(\eta\log n ) \biggr ) .
\nonumber\end{aligned}\ ] ] finally , applying the bounds ( [ eqf1 ] ) , ( [ eqgbound ] ) and ( [ eqhbound ] ) to the right of ( [ eqfinal ] ) , the proof is complete .
[ lemsch ] for two probability distributions @xmath454 and @xmath455 on @xmath456 , @xmath457 d_{{\mathrm{tv } } } ( p_1 , p_2 ) , \ ] ] if @xmath458 , where @xmath459 is the entropy of @xmath460 , @xmath461 , and @xmath462 is the total variation distance of @xmath454 and @xmath455 .
see lemma 3.1 of @xcite .
[ lemktml ] there exists a constant @xmath324 depending only on @xmath463 , such that for any @xmath85 @xmath464 the bound , see , for example , ( 27 ) in @xcite , @xmath465 where @xmath466 depends only on @xmath463 , implies the claim using @xmath467 see proof of theorem 6 in @xcite .
[ lemapprox ] let @xmath6 be a non - null stationary ergodic process with summable continuity rate .
then , for any @xmath468 and @xmath469 , @xmath220 , the empirical @xmath10-order markov estimator of the process satisfies @xmath470_1^n \bigr ) > \beta_2 { p_{\mathrm{inf}}}^{-2 } \bar{\gamma}(k ) + \frac{1}{n^{1/2-\mu } } \biggr\ } \\ & & \quad\le2 { \mathrm{e}}^{1/{\mathrm{e } } } |a|^{2+\nu\log n } \\ & & \qquad\hspace*{0pt}{}\times \exp\biggl\ { - \frac { { p_{\mathrm{inf}}}^2 } { 16 { \mathrm{e}}|a|^3 ( \alpha+ { p_{\mathrm{inf } } } ) ( \beta_1 + 1)^2 } \frac { ( n-\nu\log n ) n^{-2\nu|\log{p_{\mathrm{inf}}}| } } { ( 1+\nu\log n ) n } \\ & & \qquad\hspace*{30pt}{}\times \biggl [ n^{2\mu } - \frac{\nu|\log{p_{\mathrm{inf}}}|(\beta _ 1 + 1)^2\log n}{2 } \biggr ] \biggr\ } .\end{aligned}\ ] ] see the proof of theorem 2 and lemma 3 in @xcite .
the author would like to thank the referees for their comments that helped improving the presentation of the results and generalizing the consistency concept .
the research of the author was supported in part by nsf grant dms-09 - 06929 . | stationary ergodic processes with finite alphabets are estimated by finite memory processes from a sample , an @xmath0-length realization of the process , where the memory depth of the estimator process is also estimated from the sample using penalized maximum likelihood ( pml ) . under some assumptions on the continuity rate and the assumption of non - nullness ,
a rate of convergence in @xmath1-distance is obtained , with explicit constants .
the result requires an analysis of the divergence of pml markov order estimators for not necessarily finite memory processes .
this divergence problem is investigated in more generality for three information criteria : the bayesian information criterion with generalized penalty term yielding the pml , and the normalized maximum likelihood and the krichevsky
trofimov code lengths .
lower and upper bounds on the estimated order are obtained .
the notion of consistent markov order estimation is generalized for infinite memory processes using the concept of oracle order estimates , and generalized consistency of the pml markov order estimator is presented . |
let @xmath1 be a _ stationary ( locally finite ) poisson process _ on the space @xmath7 of _ convex bodies _ in @xmath2 , that is , on the space of compact , convex subsets of @xmath2
boolean model _ associated with @xmath1 is the stationary random closed set @xmath0 defined by @xmath8 where the poisson process @xmath1 is identified with its support .
this is a fundamental model of stochastic geometry and continuum percolation with many applications in materials science and physics @xcite .
the intersection of @xmath0 with a compact and convex set @xmath9 is a finite union of compact , convex sets , that is , an element of the _ convex ring _
@xmath10 . it is a common strategy in stochastic geometry to extract and explore local information about @xmath0 via functionals of the intersections @xmath6 .
perhaps the most prominent examples of such functionals on @xmath10 are the intrinsic volumes @xmath11 , which contain important geometric information about the sets to which they are applied .
for instance , for a set @xmath12 from the convex ring , @xmath13 is the volume , @xmath14 is half the surface area , and @xmath15 is the _
euler characteristic _ of @xmath16 ; see @xcite for more details .
the intrinsic volumes have several desirable properties .
in particular , they are _ additive _ , in the sense that @xmath17 for all @xmath18 and @xmath19 . in addition , they are translation invariant , and continuous if restricted to the space of convex bodies . for a stationary and isotropic boolean model , miles @xcite and davy @xcite obtained explicit formulas expressing the mean values @xmath20 in terms of the intensity measure of @xmath1 .
we refer to @xcite for a discussion and more recent developments related to this fundamental result . in the following , we are especially interested in second order properties and central limit theorems of the random vector @xmath21 , for a compact and convex _ observation window _
@xmath3 , but in fact we study more general additive functionals of @xmath6 , namely so called _ geometric functionals_. a functional on the convex ring will be called geometric if it is additive , translation invariant , locally bounded and measurable ( see section [ sec:3 ] for details ) . while previous contributions focus on second order properties and central limit theorems for volume and surface area , to the best of our knowledge we present here the first systematic mathematical investigation of second order properties and central limit theorems of all intrinsic volumes and more general geometric functionals of a stationary boolean model @xmath0 .
the volume functional was first studied in @xcite , while in @xcite berry - esseen bounds and large deviation inequalities were established .
the surface area was investigated in @xcite , and the results were extended in @xcite to more general functionals and point processes .
integrals over boolean models are considered in @xcite , where the volume is included as a special case and also the surface area in the latter one .
volume and surface area of a more general boolean model based on a poisson process of cylinders have been investigated in @xcite . from a geometric point of view ,
volume and surface area are rather special functionals of @xmath0 .
they arise as the restriction of deterministic measures to @xmath0 or the boundary of @xmath0 and do not involve the curvature of the ( possibly intersecting ) grains .
a different though mathematically non - rigorous treatment of second moments of curvature measures of an isotropic boolean model with an interesting application to morphological thermodynamics was presented in @xcite .
our first main aim in this paper is to use the fock space representation of poisson functionals @xcite to explore the covariance structure of geometric functionals of @xmath6 .
combined with some new integral geometric inequalities , which are derived by methods and results from convex and integral geometry , this approach appears to be perfectly tailored to our purposes . under the minimal assumption that the second moments of the intrinsic volumes of the typical grain are finite , we show that for two geometric functionals @xmath22 , @xmath23 tends to some @xmath24 as the observation window is increased in a proper way . for the case that the third moments of the intrinsic volumes of the typical grain are finite we establish a rate of this convergence in terms of the inradius of the observation window and show that it is optimal . via the fock space representation
the asymptotic covariances can be expressed as series of second moments . in the important case of intrinsic volumes of an isotropic boolean model
they can be represented in terms of curvature based moment measures of the typical grain . in particular , the covariance structure of the two - dimensional isotropic boolean model becomes surprisingly explicit . for a vector of geometric functionals of the boolean model
, it is shown that the asymptotic covariance matrix is positive definite under some additional conditions , which are for example satisfied for the intrinsic volumes .
the second order analysis is illustrated by explicit formulas for intrinsic volumes of a boolean model with deterministic spherical grains , for which our formulas reduce to three - dimensional integration of explicitly known integrands .
our second main aim is to prove univariate and multivariate central limit theorems for geometric functionals of @xmath6 . under the same second moment assumptions as for the existence of the asymptotic covariances , we prove that convergence in distribution takes place . in the univariate case we do not need to assume that the functional on the convex ring is translation invariant .
we also obtain rates of convergence under slightly stronger moment assumptions . for the multivariate central limit theorem
we provide an argument that the rate is optimal .
following common belief , we guess that our convergence rate @xmath25 for the univariate case is optimal as well . in the proofs we use the malliavin - stein method for poisson functionals that was recently developed in @xcite . in a sense
this method builds on the fock space representation and the closely related wiener - it chaos expansion of poisson functionals .
the main obstacle to the application of these results is the fact that , as a rule , geometric functionals of @xmath0 admit an infinite chaos expansion .
we can resolve this by bounding the kernels of the chaos expansion in a monotone way .
it is conceivable that the central limit theorem can also be derived with the techniques in @xcite , namely by approximation with @xmath26-dependent random fields using a truncation argument .
but such an approach would neither yield much information on the asymptotic covariance structure nor rates of convergence .
stabilization is another common approach to central limit theorems in stochastic geometry ; see e.g. @xcite .
it is unclear whether the intrinsic volumes ( other than volume or surface area ) stabilize for boolean models with unbounded grains .
moreover , in our setting we would need to control boundary effects .
this paper is organized as follows . in the second section
we briefly summarize some notation and basic facts about the boolean model . in the third section
we establish the existence of the asymptotic covariances of a vector of geometric functionals of @xmath6 and determine the rate of convergence ; see theorem [ thm : variance ] .
section [ secpos ] is devoted to the positive definiteness of the asymptotic covariance matrix ; see theorem [ posdef ] . in section [ sintegra ]
we focus on intrinsic volumes and introduce a family of curvature based moment measures of the typical grain to study infinite series of second moments arising in the fock space representation .
the main result of this section ( theorem [ tdeltaij ] ) is of some independent interest and is applied in section [ seciso ] to derive formulas for the asymptotic covariances of the intrinsic volumes of an isotropic boolean model in terms of the moment measures mentioned above ; see theorem [ tsigmaij ] .
section [ secspherical ] presents some numerical results for a boolean model with deterministic spherical grains . in section [ secclt ]
we present a general result on the normal approximation of poisson functionals .
we use this result in section [ secmclt ] to establish multivariate and univariate central limit theorems for geometric functionals of @xmath0 ; see theorem [ thm : multigeneral ] and theorem [ thm : univariat ] .
appendix a contains the proof of a crucial geometric result which is applied in section [ sintegra ] ( see lemma [ ldecomp ] ) , while appendix b provides a description of the curvature based moment measures from section [ sintegra ] in terms of mixed measures of translative integral geometry .
appendix c contains integral formulas for the exact ( non - asymptotic ) covariances of intrinsic volumes , which are rather explicit in the two - dimensional case ; see theorem [ tremainder ] and corollary [ tremainderd=2 ] .
in this section we collect a few basic facts about the stationary poisson process @xmath1 and the associated boolean model @xmath0 in euclidean space @xmath2 . all random objects occurring in this paper
are defined on an abstract probability space @xmath27 .
we consider the poisson process @xmath1 as a random element in the space @xmath28 of all locally finite counting measures on @xmath7 . here
locally finite means that every compact subset of @xmath2 is only intersected be finitely many grains .
let @xmath28 be equipped with the smallest @xmath29-field such that the mappings @xmath30 are measurable for all @xmath31 in the borel @xmath29-field of @xmath32 , the space of compact subsets of @xmath2 .
intensity measure _ @xmath33 of @xmath1 is _ translation invariant _ and therefore ( see theorem 4.1.1 in @xcite ) of the form @xmath34 where we assume that the _ intensity _ @xmath35 is positive and finite , ` @xmath36 ' denotes integration over @xmath2 with respect to @xmath37-dimensional lebesgue measure @xmath38 , and @xmath39 is a probability measure on @xmath7 satisfying @xmath40 here , as usual , @xmath41 is the _ minkowski sum _ of @xmath16 and @xmath31 , and @xmath42 , for @xmath43 .
let @xmath44 denote a _ typical grain _ , that is , a random convex set with distribution @xmath39 .
assumption is equivalent to @xmath45 the intrinsic volumes are determined by the _
steiner formula _
@xmath46 where @xmath47 is the closed unit ball centred at the origin , @xmath48 , and @xmath49 denotes the volume of the @xmath50-dimensional unit ball .
the boolean model is given by @xmath51 , where @xmath52 and @xmath53 means that @xmath54 .
recall that the mapping @xmath55 from @xmath28 to the space of all closed subsets of @xmath2 ( equipped with the fell topology ) is borel measurable ( see @xcite ) .
we can invoke the thinning property of poisson processes to assume that @xmath56 . without loss of generality
we assume that @xmath39 is concentrated on @xmath57 , the space of convex bodies with the centre of the circumscribed ball at the origin .
since the centre of the circumscribed ball of a convex body is always contained in the convex body , we have @xmath58 for all @xmath59 .
subsequently , we shall need integrability assumptions such as @xmath60 or @xmath61 we next introduce two basic characteristics of the boolean model @xmath0 . the _ volume fraction _
@xmath62^d)$ ] of @xmath0 can be expressed in the form @xmath63 the mean _ covariogram _ of the typical grain is given by @xmath64 if we assume , then @xmath65 and @xmath66 as @xmath67 , where @xmath68 denotes the euclidean norm of @xmath43 .
it is well - known ( see e.g. @xcite ) that the _ covariance _ of @xmath0 satisfies @xmath69 for @xmath70 we define by @xmath71 , @xmath43 , the set covariance function of @xmath3 . combining with fubini s theorem leads to the well - known formula @xmath72
in this paper we study random variables of the form @xmath5 , where @xmath4 is a real - valued measurable function defined on the _ convex ring _ @xmath10 whose elements are finite unions of compact , convex sets .
measurability again refers to the borel @xmath29-field generated by the fell topology ( or , equivalently , by the hausdorff metric ) .
we shall assume that @xmath4 is _ additive _ , that is , @xmath73 and @xmath74 for all @xmath18 .
we shall also assume that @xmath4 is _ translation invariant _ , that is , @xmath75 for all @xmath76 , and _ locally bounded _ in the sense that its absolute value is ( uniformly ) bounded on compact , convex sets contained in a translate of the unit cube @xmath77^d$ ] by a constant @xmath78 note that this definition simplifies in the translation - invariant case since one does not need the translations of @xmath79 . in the following , we call a functional @xmath80 _ geometric _ if it is additive , translation invariant , locally bounded , and measurable .
examples of geometric functionals are 1 .
mixed volumes ( see section 5.1 in @xcite ) of the form @xmath81,k_1,\dots , k_{d - k})$ ] , where @xmath82 and @xmath83 are fixed .
up to normalization , intrinsic volumes are obtained for @xmath84 , @xmath85 ; 2 .
integrals of surface area measures ( see sections 4.1 and 4.2 in @xcite ) of the form @xmath86 where @xmath87 is the unit sphere in @xmath2 ( the boundary of @xmath47 ) , @xmath88 is measurable and bounded , and @xmath89 ; 3 . the centred support function @xmath90 in a fixed direction @xmath91 , where @xmath92 and @xmath93 is the steiner point of @xmath16 ( see sections 1.7 and 5.4 in @xcite and lemma 6.1 in @xcite ) ; 4 .
total measures of translative integral geometry ( see section 6.4 , especially p. 234 , and p. 383 in @xcite ) . our first result deals with the asymptotic behaviour of the covariance between two geometric functionals of @xmath6 for expanding convex observation window @xmath3 .
in the following , we will consider sequences of convex bodies @xmath94 satisfying @xmath95 , where @xmath96 denotes the _ inradius _ of @xmath70 .
we describe this situation by writing @xmath97 for short . with a measurable functional @xmath98 , we associate another measurable function @xmath99 by @xmath100 if @xmath101 . under assumption
, it follows from that @xmath102 is well defined for a geometric functional @xmath4 .
[ thm : variance ] let @xmath103 and @xmath104 be geometric functionals . if is satisfied , then the limit @xmath105 exists and is given by @xmath106 if holds , then there is a constant @xmath107 such that @xmath108 for @xmath70 with @xmath109 . from the proof of theorem [ thm : variance ]
it also follows that the right - hand side of is still convergent if we take the absolute values of the integrands .
we start with some preparations .
our main probabilistic tool is the following fock space representation of poisson functionals , derived in @xcite . for any measurable @xmath110 and @xmath111 , the function @xmath112 is defined by @xmath113 where @xmath114 is the dirac measure located at @xmath16 .
the _ difference operator _ @xmath115 and its iterations play a central role in the analysis of poisson processes . for @xmath116 and @xmath117
we define a function @xmath118 inductively by @xmath119 where @xmath120 .
note that @xmath121 where @xmath122 denotes the number of elements of @xmath123 .
this shows that the operator @xmath124 is symmetric in @xmath125 , and that @xmath126 is measurable . from (
* theorem 1.1 ) we obtain for any measurable @xmath127 satisfying @xmath128 and @xmath129 that @xmath130 in particular , this implies that the right - hand side of still converges if we take the absolute values of the expectations . for given @xmath70 and a functional @xmath98 we shall apply to functions @xmath131
defined by @xmath132 [ lem : kernels ] let @xmath98 be additive .
then , for @xmath133 , @xmath134 , and @xmath135 , @xmath136 this easily follows by induction . in the following ,
we write @xmath137 for constants which depend only on the dimension , the intensity , and the distribution of the typical grain . [
lem : boundintrinsic ] let @xmath4 be an additive , locally bounded , and measurable functional and assume that is satisfied . then , for all @xmath133 , @xmath138 , and @xmath70 , @xmath139 and @xmath140 only depending on @xmath4 .
moreover , for any @xmath141 , @xmath142 we start with proving that @xmath143 for @xmath141 .
since is obviously true for @xmath144 , we assume @xmath145 in the following .
we define @xmath146 , where we recall that @xmath147^d$ ] . by the inclusion - exclusion formula for additive functionals ( see e.g. ( * ? ? ?
* ( 3.4.3 ) ) ) , we have @xmath148 for each non - empty subset @xmath149 , we fix some cube @xmath150 .
let @xmath151 denote the number of all particles in @xmath1 intersecting @xmath152 , and let @xmath153 denote the particles hitting @xmath154 .
then , for @xmath155 , assumption yields that @xmath156 by the inclusion - exclusion formula and taking into account that @xmath73 , we get @xmath157 the cubes in @xmath158 form a grid , hence @xmath159 by stationarity of @xmath1 , we have @xmath160 , and thus @xmath161 here we have used that for @xmath162 @xmath163 where @xmath164 is the reflection of @xmath31 in the origin . since @xmath165 , steiner s formula yields @xmath166 and the definition of @xmath102 in shows . for @xmath141
, we can argue as in the derivation of , and then use , to get @xmath167 with lemma [ lem : kernels ] yields . in order to show that @xmath168 is square integrable
, we first derive an upper bound for @xmath169 let @xmath170 with @xmath171 . then , using the inclusion - exclusion formula for additive functionals and , we get @xmath172 and hence @xmath173 again by the inclusion - exclusion formula , we have @xmath174\ , 4^{|\mathcal{q}(a)|}m(\psi)^2<\infty,\ ] ] which concludes the proof .
[ lem : translativeintegral ] there is a constant @xmath175 , depending only on the dimension , such that @xmath176 for all @xmath177 and @xmath178 . using the same notation as in the proof of lemma 3.3 and the fact that @xmath179 is increasing and translation invariant
, we obtain that @xmath180 let @xmath181 denote a ball of radius @xmath182 .
then @xmath183 where @xmath184 is a constant depending only on the dimension and we have used the kinematic formula ( see ( * ? ? ? * theorem 5.1.3 ) ) and the rotation invariance of @xmath181 . on the other hand ,
it was shown in the proof of lemma 3.3 that @xmath185 combining the preceding inequalities , we obtain the assertion of the lemma .
[ lem : kinematic ] for @xmath141 and @xmath133 , @xmath186 where @xmath187 with @xmath188 as in lemma [ lem : translativeintegral ] . in the following calculation and also later , we use the convention @xmath189 .
we apply and to get @xmath190 by iterating this step @xmath191 more times , we obtain the assertion .
[ bounds ] there is a constant @xmath192 , depending only on the dimension , such that @xmath193 for @xmath194 .
let @xmath195 .
then we have @xmath196 let @xmath197 for @xmath43 and a closed set @xmath198 , and let @xmath199 . then @xmath200 to see this , let @xmath201 denote the metric projection to @xmath3 and consider the map @xmath202 , @xmath203 . let @xmath204 and choose a point @xmath205 such that @xmath206 . using that @xmath207 is an outer normal of @xmath3 at @xmath208 it is easy to see that @xmath209 .
hence @xmath210 . since the metric projection is @xmath211-lipschitz it is not hard to prove that @xmath212 has the same property . therefore follows .
this yields that @xmath213 where steiner s formula was used .
[ lem : boundinradius ] let @xmath70 such that @xmath214 and let @xmath215 . then @xmath216 steiner s formula and the fact that @xmath217 , for @xmath218 , imply that @xmath219 which concludes the proof .
let @xmath70 .
in order to compute the numerator in we shall apply with @xmath220 and @xmath221 .
from we conclude that indeed @xmath222 and @xmath223 .
since @xmath0 is stationary , the translation invariance of a functional @xmath224 implies that @xmath99 defined by is translation invariant as well . from
we get @xmath225 for @xmath70 and @xmath133 , we define @xmath226 by @xmath227 and @xmath228 by @xmath229 our aim is to prove that @xmath230 as @xmath97 . since we want to apply the dominated convergence theorem , we provide an upper bound for @xmath231 , which is independent of @xmath3 .
it follows from in lemma [ lem : boundintrinsic ] , the translation invariance of @xmath232 and @xmath233 , and the monotonicity of the intrinsic volumes that @xmath234 combining this estimate with lemma [ lem : translativeintegral ] and lemma [ lem : kinematic ] , we get @xmath235 by the right - hand side of is integrable .
moreover , lemma [ lem : boundinradius ] shows that it is uniformly bounded for @xmath70 with @xmath109 , and the same holds if we sum over all @xmath133 .
next we bound @xmath236 . by using the translation invariance of @xmath237 and @xmath233 , we have @xmath238 note that the integrand is zero if @xmath239 and @xmath240 .
the same holds for the case that @xmath241 and @xmath242 .
this means that the integrand can be only non - zero if @xmath243 . on the other hand ,
the integrand is always bounded by @xmath244 where we have used lemma [ lem : boundintrinsic ] and the monotonicity of the intrinsic volumes .
hence , we obtain that @xmath245 where we have used the fact that @xmath232 is increasing and the translation invariance of @xmath232 and @xmath233 in the last step .
now lemma [ lem : kinematic ] and lemma [ bounds ] yield that @xmath246 together with lemma [ lem : boundinradius ] this shows that for any @xmath70 with @xmath109 , @xmath247 where @xmath248 depends only on @xmath249 .
therefore , for any @xmath70 with @xmath109 , @xmath250 thus an application of the dominated convergence theorem yields the convergence result stated in the theorem . under the stronger moment assumption
, the rate of convergence in follows from by carrying out the integration with respect to @xmath16 .
if the geometric functional is the volume , the asymptotic variance has a significantly easier representation than in , namely @xmath251 this follows from an application of the dominated convergence theorem to the exact variance formula .
the inequalities @xmath252 , @xmath253 , and @xmath254 imply that @xmath255 together with @xmath256 this means that @xmath257 is integrable and is an upper bound for @xmath258 . now
the observation that @xmath259 as @xmath97 for any @xmath260 ( this follows from @xmath261 , , steiner s formula , and lemma [ lem : boundinradius ] ) yields
. in section [ seciso ] formulas as are derived for the other intrinsic volumes .
the following proposition shows that the rate of convergence stated in theorem [ thm : variance ] is optimal .
[ prop : optimalrate ] assume that is satisfied and that the typical grain is full - dimensional with positive probability .
then there is a constant @xmath262 such that @xmath263 for @xmath70 with @xmath109 .
recall from the proof of lemma [ bounds ] that @xmath264 , for @xmath265 and a closed set @xmath198 , and @xmath266 for @xmath267 .
for @xmath268 we define @xmath269 .
then @xmath270 is convex , the boundary of @xmath271 is @xmath272 , and @xmath273 is strictly decreasing with respect to set inclusion , for @xmath274 $ ] .
it follows from and that @xmath275 since the typical grain is full - dimensional with positive probability , there are constants @xmath276 and @xmath277 such that @xmath278 for all @xmath279 .
this means that @xmath280 denoting by @xmath281 the closed ball with centre @xmath282 and radius @xmath283 , we have @xmath284 using the fact that @xmath285 for @xmath286 with @xmath287 , we obtain @xmath288 it follows from lemma 3.2.34 in @xcite that @xmath289 for @xmath290 $ ] .
the discussion at the beginning of this proof implies that @xmath291 is strictly decreasing on @xmath292 $ ] .
together with @xmath293 we get for @xmath294 that @xmath295 combining this with and concludes the proof .
let @xmath296 be geometric functionals on @xmath297 .
we assume that @xmath298 , for @xmath299 , is positively homogeneous of degree @xmath300 and @xmath301 , with a constant @xmath302 which only depends on @xmath298 . by theorem [ thm :
variance ] , for @xmath303 , the asymptotic covariances @xmath304 exist under the assumption .
the following theorem shows that the asymptotic covariance matrix is positive definite .
in particular , the result applies to the intrinsic volumes @xmath11 , which also means that their asymptotic variances are strictly positive .
[ posdef ] let the preceding assumptions and be satisfied .
moreover assume that the typical grain @xmath44 has non - empty interior with positive probability .
then the covariance matrix @xmath305 is positive definite . for a vector @xmath306
we have @xmath307 since each summand is non - negative , the matrix @xmath308 is positive definite if we can prove that one summand is greater than zero for a given @xmath309 .
specifically , under the given assumptions we shall show that the summand obtained for @xmath310 is positive . in order to show this
, we shall prove that for @xmath311 in the support of @xmath39 and having non - empty interiors , there is a set of translation vectors @xmath312 of positive @xmath313 measure ( recall that @xmath38 denotes @xmath37-dimensional lebesgue measure ) for which @xmath314 for the rest of the proof , we argue with a convex body @xmath315 .
properties which will be required of @xmath316 will be provided by an application of lemma [ lem : intersection ] and @xmath316 of the form @xmath317 , for a set of translation vectors @xmath312 of positive @xmath313 measure .
this will finally prove the preceding assertion and thus the theorem . for @xmath315 ,
let @xmath318 be the number of grains of @xmath1 that intersect @xmath316 , but do not contain it , and let @xmath319 be the number of grains of @xmath1 that contain @xmath316 . then @xmath318 and @xmath319 are independent , poisson distributed random variables with parameters @xmath320 if @xmath321 , then @xmath322 , and therefore @xmath323 if @xmath324 , then @xmath325 , and hence @xmath326 this leads to @xmath327 = -\exp(-s_1(l)-s_2(l))\psi_k(l)+r_k(l),\ ] ] where @xmath328 next we bound @xmath329 from above .
so assume that @xmath330 .
let @xmath331 denote the grains of @xmath1 which hit @xmath316 , but do not contain @xmath316 . with the definition of @xmath332 from , we obtain from that @xmath333 in the following , let @xmath334 stand for the radius of the circumscribed ball of @xmath111 . for @xmath141 with @xmath335 ,
let @xmath336 be the centre of the circumball of @xmath337 , hence @xmath338 .
since @xmath298 is locally bounded and homogeneous of degree @xmath300 , we get @xmath339 and hence @xmath340 thus , in the present case , we have @xmath341 hence the remainder term can be bounded from above by @xmath342 next we derive an upper bound for @xmath343 . by definition and the reflection invariance of lebesgue measure
, we have @xmath344 to bound the inner integral from above , we can assume that @xmath345 , by the translation invariance of lebesgue measure . if the integrand is non - zero , then @xmath346 or @xmath347 .
then the inequality implies that the inner integral is bounded from above by @xmath348 .
hence , if @xmath349 , steiner s formula and our moment assumption yield that @xmath350 where as before @xmath351 denotes a constant depending on @xmath352 . hence ,
if @xmath353 is sufficiently small , then @xmath354 , and thus @xmath355 we also have from that latexmath:[\[\label{eqc } hence , if @xmath353 is sufficiently small , we deduce from , , and that @xmath357 is a constant depending on @xmath358 .
in addition , latexmath:[\[\label{eqe } if @xmath354 , with @xmath302 as in .
let @xmath360 be the smallest @xmath299 such that @xmath361 . then
, if @xmath353 is sufficiently small , we get @xmath362 where we used and , for @xmath363 , and for @xmath364 . here
we denote by @xmath365 a constant which depends on @xmath366 .
the lower bound thus obtained is positive if @xmath353 is sufficiently small and @xmath367 , for some constant @xmath368 . the proof is completed by an application of lemma [ lem : intersection ] below .
the following lemma on the ratio of circumradius and inradius of translates of convex bodies is a key argument in the proof of theorem [ posdef ] .
[ lem : intersection ] for all @xmath369 with non - empty interior there is a constant @xmath370 such that @xmath371 for all @xmath372 .
let @xmath373 be unit vectors whose endpoints are the vertices of a regular simplex .
for @xmath374 let @xmath375 be a point in the boundary of @xmath376 which has @xmath377 as an exterior normal vector .
the support cone @xmath378 of @xmath376 at @xmath375 ( cf .
@xcite ) then satisfies @xmath379 where @xmath380 is the supporting half - space of @xmath376 with exterior unit normal @xmath377 and @xmath381 denotes the closure . by
* theorem 12.2.2 ) it follows that @xmath382 in the topology of closed convergence as @xmath383 .
moreover , since @xmath311 have non - empty interiors , there are vectors @xmath384 such that the origin is an interior point of @xmath385 and the circumradius of the intersection on the right - hand side is less than @xmath211 ( say ) .
then ( * ? ? ?
* theorem 1.8.8 ) and ( * ? ? ?
* theorem 12.3.3 ) imply that @xmath386 where @xmath387 and the convergence is with respect to the hausdorff distance .
since the inradius and the circumradius of the intersection of translates of convex bodies are continuous with respect to the translations as long as the intersection has non - empty interior , there is some @xmath388 such that the ratio between inradius and circumradius of @xmath389 is close to the corresponding ratio of @xmath390 , for @xmath391 , and therefore also @xmath392 with a constant @xmath393 which depends only on @xmath311 .
moreover , for @xmath394 we have @xmath395 and thus @xmath396 therefore , if @xmath397 the proof of the lemma is completed by remarking that the intersections are continuous with respect to translations as long as the intersection has non - empty interior and by using the translation invariance of lebesgue measure .
clearly , this proves the lemma for all @xmath372 .
[ rem4.1]so far we established the positive definiteness of the asymptotic covariance matrix of @xmath398 functionals @xmath298 , @xmath299 , where @xmath298 was assumed to be homogeneous of degree @xmath300 .
the choice of @xmath398 functionals of the given degrees of homogeneity is motivated by the example of the intrinsic volumes .
however , it should be noted that the preceding argument immediately extends to any finite number of functionals @xmath399 having different non - negative degrees of homogeneity @xmath400 and satisfying a lower bound corresponding to , that is , @xmath401 for all @xmath111 .
we shall see in the next section that in the particularly important case of intrinsic volumes and under the assumption of isotropy the asymptotic covariances of theorem [ thm : variance ] can be expressed in terms of the numbers @xmath402 in this section we study these numbers _ without isotropy assumption _ on @xmath0 .
the results are of independent interest . for @xmath70 and @xmath403 , we define @xmath404 which is a finite window version of @xmath405 .
the numbers @xmath406 are further studied in appendix c. the relationship between and is given in the next corollary .
[ cor : alimit ] let @xmath403 .
if is satisfied , then @xmath407 and @xmath408 if is satisfied , then there is a constant @xmath409 such that @xmath410 for @xmath70 with @xmath109 .
this can be proved in a similar way as theorem [ thm : variance ] .
the previous corollary describes @xmath405 as the limit of @xmath411 for observation windows with @xmath97 .
it is , however , more convenient to work with the series representation .
we shall see that this series can be expressed in terms of a finite family of ( curvature ) measures @xmath412 to be introduced below . for @xmath413 and @xmath414 , we let @xmath415 denote the @xmath416-th curvature measure of @xmath16 ( see @xcite ) . in particular
, @xmath417 is the restriction of lebesgue measure to @xmath16 while @xmath418 is half the @xmath419-dimensional hausdorff measure restricted to the boundary of @xmath16 ( if the affine hull of @xmath16 has full dimension ) .
furthermore , @xmath420 for all @xmath421 . for @xmath422 , @xmath133 , and @xmath423 we define @xmath424 since @xmath425 , @xmath422 , is concentrated on the boundary @xmath426 of @xmath427 , this definition is consistent with the case @xmath428 . for @xmath429 and @xmath430 , we define a measure @xmath431 on @xmath2 by @xmath432 with the appropriate interpretation of the case @xmath433 . for @xmath434 , @xmath430 , @xmath435 , and
@xmath436 we define a measure @xmath437 on @xmath2 by @xmath438 where @xmath439 denotes the interior of @xmath414 and with the appropriate interpretation of the cases @xmath440 or @xmath441 .
let @xmath442 and , for @xmath422 , @xmath443 moreover , we define @xmath444 @xmath445 for @xmath422 , and @xmath446 , where @xmath447 is the dirac measure concentrated at the origin and @xmath448 is the mean covariogram of the typcial grain as defined in .
subsequently , we assume that @xmath449 that is , the typical grain almost surely has non - empty interior .
[ tdeltaij ] assume that and are satisfied .
then the measures @xmath412 are all finite
. moreover , the limits are given by @xmath450 for @xmath451 the result remains true without assumption .
in particular , we thus have @xmath452 @xmath453 , and @xmath454 for @xmath422 .
the proof of theorem [ tdeltaij ] is based on the following geometric auxiliary result , whose proof will be given in appendix a. here we use the abbreviation @xmath455=\{1,\hdots , n\}$ ] .
[ ldecomp ] let @xmath456 and @xmath457 .
assume that is satisfied and @xmath458 has non - empty interior
. then @xmath459\\ |i|=k } } \phi_i\bigg(\bigcap_{r\in i}k_r;\cdot\cap \bigcap_{r\in i}\partial k_r \cap \bigcap_{s\notin i}k^\circ_s\bigg)\end{aligned}\ ] ] for @xmath460-a.e .
@xmath461 .
we start with showing that the measures @xmath412 are finite .
let @xmath434 , @xmath430 , @xmath435 , and @xmath436 .
then @xmath462 for @xmath463 the right - hand side is finite because of assumption .
otherwise we obtain by lemma [ lem : kinematic ] and lemma [ lem : translativeintegral ] that @xmath464 now it follows from that the right - hand side is finite .
similar ( but easier ) arguments show that the other measures are also finite . to prove that the series is given by , we distinguish different cases and start with @xmath465 .
then we have @xmath466 for @xmath467 and @xmath451 we get @xmath468 and hence @xmath469 this and the preceding calculation remain true without assumption , the reason being that the integration in can effectively be restricted to full - dimensional grains . next we turn to @xmath467 and @xmath470 . then , using @xmath471 , for @xmath472 , and lemma [ ldecomp ] , we get @xmath473\\ |j|=l}}\iint \phi_j\left(\bigcap_{r\in j}k_r;\bigcap_{r\in j}\partial k_r\cap\bigcap_{s\notin j}k_s^\circ\right ) \,\lambda^{n-1}(d(k_2,\ldots , k_n ) ) \ , \mathbb{q}(dk_1)\\ & = \gamma \sum^\infty_{n=1}\frac{1}{n ! } \sum_{l=1}^{d - j}\sum_{\substack{j\subset [ n]\\ |j|=l}}\iiint \i\{z\in \cap_{s\notin j}k_s^\circ\}\ , \phi_j(k_j;dz)\,\lambda^{n-1}(d(k_2,\ldots , k_n))\ , \mathbb{q}(dk_1),\end{aligned}\ ] ] where @xmath474 for @xmath475 ( see ) . at this stage and
also later we use the covariance property @xmath476 which holds for all measurable @xmath477 $ ] .
this follows from the definition and ( * ? ? ?
* theorem 14.2.2 ) . using and then the invariance of @xmath233 under translations , it is easy to check that , for instance
, @xmath478 from such symmetry relations we deduce that @xmath479 next we address the case @xmath429 and @xmath451 . using again lemma [ ldecomp ] and a symmetry argument ( as above )
, we obtain @xmath480 \\ |i|=k}}\iiint \i\{y\in \cap_{r\notin i}k_r^\circ\}v_d(k_1\cap\cdots\cap k_n)\\ & \qquad\qquad\phi_i(k_i;dy)\,\lambda^{n-1}(d(k_2,\ldots , k_n))\ , \mathbb{q}(dk_1)\\ & = \gamma \sum^\infty_{n=1}\frac{1}{n ! } \sum_{k=1}^{d - i}\binom{n}{k}\iiiint \i\{y\in k_{k+1}^\circ\cap\ldots\cap k_n^\circ\ } \i\{z\in k_1\cap\cdots\cap k_n\}\\ & \qquad\qquad dz\ , \phi_i(k_1,\ldots , k_k;dy)\ , \lambda^{n-1}(d(k_2,\ldots , k_n))\ , \mathbb{q}(dk_1).\end{aligned}\ ] ] then we interchange the order of summation to get @xmath481 which yields that @xmath482 finally , we consider the case where @xmath434 . again by lemma [ ldecomp ] we get @xmath483\\ |i|=k}}\sum_{\substack{j\subset [ n]\\ |j|=l}}\iiiint \i\{y\in \cap_{r\notin i}k_r^\circ , z\in \cap_{s\notin j}k_s^\circ\}\\ & \qquad\qquad\phi_i(k_i;dy)\,\phi_j(k_j;dz)\,\lambda^{n-1}(d(k_2,\ldots , k_n))\ , \mathbb{q}(dk_1)\\ & = \gamma \sum^\infty_{n=1}\frac{1}{n!}\sum_{k=1}^{d - i } \sum_{l=1}^{d - j}\sum_{m=0}^{k\wedge l } \sum_{\substack{i , j\subset [ n]\\ |i|=k,|j|=l,|i\cap j|=m } } \iiiint \i\{y\in \cap_{r\notin i}k_r^\circ , z\in \cap_{s\notin j}k_s^\circ\}\\ & \qquad\qquad\phi_i(k_i;dy)\,\phi_j(k_j;dz)\,\lambda^{n-1}(d(k_2,\ldots , k_n))\ , \mathbb{q}(dk_1).\end{aligned}\ ] ] a symmetry argument shows ( as before ) that for each choice of @xmath484 such that @xmath485 , @xmath486 and @xmath487 , the preceding integral has the same value
. there are @xmath488 possible choices of @xmath484 with these properties .
thus we obtain @xmath489 and hence @xmath490 this concludes the proof of the theorem .
the measures @xmath491 , @xmath492 , can be expressed in terms of the mixed moment measures @xmath493 and the functions @xmath494 , defined by @xmath495 [ limxed ] assume that is satisfied . then , for any @xmath496 , @xmath497 equations and follow directly from the definitions , while follows from an easy calculation using the covariance property .
lemma [ limxed ] implies in particular that @xmath498 these formulas will be used in the next section .
in this section we assume that the typical grain is _ isotropic _ , that is , its distribution @xmath39 is invariant under rotations and that the moment assumption is satisfied .
our aim is to derive more explicit formulas for the asymptotic covariances @xmath499 confer the statement of theorem [ thm : variance ] . using the iterated version of the local kinematic formula ( * ? ? ?
* theorem 5.3.2 ) , which is obtained by combining ( * ? ? ? * theorem 6.4.1 , ( b ) and ( 6.15 ) ) and ( * ? ? ?
* theorem 6.4.2 , ( 6.20 ) ) , we obtain for @xmath422 and @xmath435 that @xmath500 where , as in ( * ? ? ?
* ( 5.4 ) ) , @xmath501 combining this with and theorem [ tdeltaij ] , and under the assumption , we get @xmath502 where @xmath503 ( a multivariate polynomial on @xmath504 of degree @xmath37 ) is defined by @xmath505 the following main result of this section shows that the asymptotic covariances are linear combinations of the numbers @xmath405 given by . to describe the coefficients , we define for any @xmath422 and @xmath506 a polynomial @xmath507 on @xmath504 of degree @xmath508 by @xmath509 and complement this definition by @xmath510 .
[ tsigmaij ] assume that the typical grain is isotropic and suppose that holds .
then @xmath511 for all @xmath403 .
the formula preceding theorem 9.1.4 in @xcite is the finite volume version of the fundamental result of @xcite and @xcite on the densities of intrinsic volumes . using this result
, we obtain for all @xmath512 and @xmath141 that @xmath513 for @xmath514 , equation is a direct consequence of stationarity and the definition @xmath515 . using this formula in
, we obtain the assertion from .
[ surfvol ] assume that is satisfied .
then , for @xmath516 , the assertions of theorem [ tsigmaij ] remain true in the general stationary case ( without isotropy assumption ) .
moreover , @xmath517 if , in addition , holds , then @xmath518 the formula preceding theorem 9.1.4 in @xcite does not require isotropy for @xmath519 .
therefore , for @xmath516 , the proof of theorem [ tsigmaij ] applies without this assumption . by definition , @xmath520 and @xmath521 .
therefore we obtain from theorem [ tsigmaij ] that @xmath522 , @xmath523 inserting first , , and then and , we obtain the result . together with corollary [ surfvol ]
the next corollary provides rather explicit formulas for the asymptotic covariance in the two - dimensional isotropic case .
[ 2d ] let @xmath524 , assume that the typical grain is isotropic , and suppose that and are satisfied .
then @xmath525 where @xmath526 the formula for @xmath527 remains true without assumption .
we have @xmath528 , @xmath529 , @xmath530 , @xmath531 , @xmath532 , and @xmath533 .
moreover , we have @xmath534 and @xmath535 . using , theorem [ tdeltaij ] , and lemma [ limxed ] , we obtain @xmath536 the result follows by substituting these expressions into theorem [ tsigmaij ] .
the proof of theorem [ tsigmaij ] also yields the following non - asymptotic result for which definition should be recalled .
[ tsigmaijw ] assume that the typical grain is isotropic and that holds .
let @xmath70 and @xmath403 .
then @xmath537 further discussion of the case @xmath524 will be provided in appendix c.
in this section we show how some of the formulas of section [ seciso ] can be used to determine explicitly the covariances of a stationary and isotropic boolean model whose typical grain is the unit ball @xmath47 . in this particular case , we get from corollary [ surfvol ] that @xmath538,\ ] ] where @xmath539 and @xmath540 denotes the @xmath416-dimensional hausdorff measure .
clearly , @xmath541 , for @xmath253 and @xmath542 , is independent of the choice of the unit vector @xmath543 and @xmath544.\ ] ] introducing polar coordinates , we get @xmath545 where @xmath546 . on the other hand , for an arbitrary unit vector @xmath542 , by the rotation invariance of @xmath47 we get @xmath547
we parametrize @xmath208 in the form @xmath548,\ , s\in [ 0,1],\ , w\in v^\perp\cap s^{d-1},\ ] ] and hence we obtain @xmath549 therefore we have @xmath550 which shows that the sign of the covariance @xmath551 is completely determined by the sign of the function @xmath552 .
it is preferable to plot the covariances as functions of the intensity .
here we have @xmath553 figure [ covcolorall ] shows the result for various dimensions . for @xmath524 , , @xmath554,width=302
] next we determine the correlation coefficient @xmath555 , as a function of the intensity @xmath35 .
for this we also have to determine explicitly @xmath556 and @xmath557 , which requires some further calculations .
first , we have @xmath558 hence @xmath559 second , @xmath560.\end{aligned}\ ] ] in order to determine @xmath561 explicitly , we introduce the surface area of a spherical cap of height @xmath562 , denoted by @xmath563 , for @xmath564 $ ] , that is , @xmath565 therefore @xmath566 for @xmath567 .
let @xmath568 be fixed . then , arguing as in the derivation of , we obtain @xmath569 furthermore , we have @xmath570 finally , since @xmath571 we get ( with an arbitrary unit vector @xmath572 ) @xmath573\right)\right)\sin^{d-2}\theta\ , \ , d\theta\ , \mathcal{h}^{d-2}(dv)\\ & = \frac{d\kappa_d(d-1)\kappa_{d-1}}{4}\int_0^\pi \sin^{d-2}\theta\exp\left(\gamma\bar{c}_d\left(\sqrt{2(1-\cos\theta)}\right)\right)\ , d\theta\\ & = \frac{d\kappa_d(d-1)\kappa_{d-1}}{4}\int_{0}^2 \sqrt{s(2-s)}^{d-3 } \exp\left(\gamma\bar{c}_d\left(\sqrt{2(2-s)}\right)\right)\ , d s\\ & = : \frac{d\kappa_d(d-1)\kappa_{d-1}}{4}k_d(\gamma).\end{aligned}\ ] ] hence , we have @xmath574 this finally implies that @xmath575 from these considerations we also deduce the plausible fact that @xmath576 which is confirmed by our numerical calculations
. plots of @xmath577 for @xmath578 are shown in figure [ figcorr2 ] . for @xmath578,width=302 ] in a similar way , the formulas from corollary [ 2d ] can be specified in the case of a planar boolean model with the unit circle as deterministic typical grain
. then we have @xmath579 and , for instance , @xmath580 where @xmath581 .
moreover , @xmath582 @xmath583 the variances and covariances as well as the correlation functions for the planar case are plotted in figure [ figcorr3 ] and figure [ figcorr4 ] .
the unique zero @xmath584 of @xmath585 satisfies @xmath586 $ ] , whereas @xmath527 has two zeros @xmath587 with @xmath588 $ ] and @xmath589 $ ] .
the unique zero @xmath590 of @xmath591 satisfies @xmath592 $ ] ; see also figure [ covcolorall ] .
a local minimum of @xmath593 is in @xmath594 $ ] with value @xmath595 , @xmath596 has a local minimum in @xmath597 $ ] with value @xmath598 .
a local maximum of @xmath585 is in @xmath599 $ ] with value @xmath600 , a local maximum of @xmath596 is in @xmath601 $ ] with value @xmath602 .
a global maximum of @xmath585 is in @xmath603 $ ] with value @xmath604 , and a global minimum is in @xmath605 $ ] with value @xmath606 .
close to the global maximum of @xmath585 is the global maximum of @xmath527 , which is in @xmath607 $ ] with value @xmath608 .
moreover , @xmath527 has a global minimum in @xmath609 $ ] with value @xmath610 , and a local maximum in @xmath611 $ ] with value @xmath612 .
furthermore , @xmath613 has a global minimum in @xmath614 $ ] with value @xmath615 , and a local maximum in @xmath616 $ ] with value @xmath617 . for large intensities , @xmath618 and @xmath619 and @xmath620 .
, width=302 ] , width=302 ]
in this section we prepare the central limit theorems for geometric functionals of a boolean model by proving a general result on the normal approximation of poisson functionals .
our approach is based on recent results from @xcite . throughout this section ,
let @xmath1 be a poisson process on a measurable space @xmath621 with a @xmath29-finite intensity measure @xmath622 , see ( * ? ? ?
* chapter 12 ) .
consider a @xmath623$]-valued random variable @xmath624 such that @xmath625 and @xmath626 @xmath627-a.s .
for some measurable @xmath110 .
any such @xmath628 is called _ representative _ of the _ poisson functional _ @xmath624 . if @xmath628 is a ( fixed ) representative of @xmath624 we define @xmath629 where @xmath630 is the @xmath50-th iterated difference operator used in section [ sec:3 ] .
if @xmath631 is another representative of @xmath624 , then the multivariate mecke equation ( see e.g. ( * ? ?
* ( 2.10 ) ) ) implies that @xmath632 @xmath627-a.s . and for @xmath633-a.e .
let @xmath635 denote the space of all poisson functionals @xmath624 such that @xmath636 .
for @xmath637 we define @xmath638 by @xmath639 it was shown in @xcite that @xmath640 belongs to the space @xmath641 of symmetric functions on @xmath642 that are square - integrable with respect to @xmath633 . now the fock space representation ( see @xcite ) tells us that @xmath643 where @xmath644 denotes the norm in @xmath645 .
moreover , it is known from @xcite that @xmath624 has the representation @xmath646 where @xmath647 stands for the @xmath50-th multiple wiener - it integral , and the right - hand side converges in @xmath648 .
the identity is called wiener - it chaos expansion of @xmath624 .
the multiple wiener - it integrals are defined for square integrable symmetric functions and are orthogonal in the sense that @xmath649 for @xmath650 , @xmath651 , and @xmath652 , where @xmath653 denotes the scalar product in @xmath645 . if the condition @xmath654 is satisfied , the difference operator has the representation @xmath655 @xmath656-a.s . for @xmath622-a.e .
@xmath657 ( see e.g. ( * ? ? ? * theorem 3.3 ) ) . from now on ,
we write @xmath658 if @xmath637 satisfies . for a poisson functional @xmath637 such that @xmath659 the ornstein - uhlenbeck generator is given by @xmath660 and
its pseudo - inverse is @xmath661 for @xmath637 .
these operators together with the difference operator and the skorohod integral , which is not used in this paper , are called malliavin operators . combining and
, we see that @xmath662 @xmath656-a.s . for @xmath622-a.e .
more details on the wiener - it chaos expansion and the malliavin operators can be found in @xcite and the references therein . in @xcite ,
the malliavin operators and stein s method are combined to derive bounds for the normal approximation of poisson functionals . in the following , we evaluate bounds obtained by this technique , which is called malliavin - stein method .
to measure the distance between two real - valued random variables @xmath663 , we use the wasserstein distance that is given by @xmath664 here , @xmath665 stands for the set of all functions @xmath666 with a lipschitz constant less than or equal to one . for two @xmath26-dimensional random vectors
@xmath663 we define @xmath667 where @xmath668 is the set of all three times continuously differentiable functions @xmath669 such that @xmath670 convergence in the wasserstein distance or in the @xmath671-distance implies convergence in distribution .
in the following , we establish an upper bound for the @xmath671-distance between a gaussian random vector and a random vector @xmath672 of poisson functionals @xmath673 .
each of these components has a wiener - it chaos expansion @xmath674 with @xmath675 , @xmath133 .
we also state a bound for the wasserstein distance between the normalization of a poisson functional @xmath624 and a standard gaussian random variable .
we need to introduce some notation .
consider functions @xmath676 and @xmath677 , where @xmath678 . the tensor product @xmath679 is the function on @xmath680 which maps each @xmath681 to @xmath682 .
this definition can be iterated in the obvious way .
fix two integers @xmath683 and consider functions @xmath684 and @xmath685 .
let @xmath29 be a partition of @xmath686 and let @xmath687 be the number of blocks ( i.e. , subsets of the partition ) of @xmath29 .
the function @xmath688 is defined by replacing all variables whose indices belong to the same block of @xmath29 by a new common variable .
let @xmath689 be the partition of @xmath690 into the sets @xmath691 , @xmath692 , @xmath693 , and @xmath694 .
let @xmath695 be the set of all partitions @xmath29 of @xmath690 such that @xmath696 for all @xmath697 and all @xmath698 .
by @xmath699 we denote the set of all partitions @xmath700 such that * @xmath701 or @xmath702 ; * each block of @xmath29 has at least two elements ; * for every partition of @xmath703 in two disjoint non - empty sets @xmath704 there are @xmath705 , @xmath706 such that @xmath707 and @xmath708 are both intersected by one block of @xmath29 .
let @xmath709 ( resp .
@xmath710 ) be the set of all partitions @xmath711 such that @xmath701 ( resp .
@xmath712 ) .
now we are able to state the main result of this section : [ thm : abstract ] assume that @xmath713 and @xmath714 for all @xmath700 , @xmath715 , and @xmath716 .
assume that that there are @xmath717 and @xmath718 such that @xmath719 for all @xmath711 , @xmath720 , and @xmath721 .
let @xmath722 and let @xmath723 be a centred gaussian random vector with a given positive semidefinite covariance matrix @xmath724 .
then @xmath725 in the univariate case , we have the following result for the wasserstein distance : [ corol : univariat ] let @xmath658 such that @xmath726 and the assumptions and are satisfied and let @xmath723 be a standard gaussian random variable .
then @xmath727 we prepare the proof of theorem [ thm : abstract ] by two lemmas and a proposition .
[ lem : rij ] let @xmath683 , @xmath728 , @xmath729 , and assume that @xmath730 then @xmath731 combining the formulas @xmath732 and @xmath733 with theorem 3.1 in @xcite ( see also @xcite ) proves the first equation .
the second identity is a consequence of @xmath734 and , again , theorem 3.1 in @xcite .
[ prop : abstractbound ] let @xmath735 be such that holds .
let @xmath736 and let @xmath723 be a centred gaussian random vector with a given positive semidefinite covariance matrix @xmath724 .
then @xmath737 from ( * ? ? ?
* theorem 4.2 ) it is known that @xmath738 we bound the two summands on the above right - hand side separately .
for the first one we have @xmath739 from and and the covariance version of ( see ) we obtain that @xmath740 where @xmath741 using the triangle inequality and , we obtain @xmath742 where we have applied in lemma [ lem : rij ] to get the final inequality . by jensen s inequality and the definitions of the malliavin operators , we obtain for the second summand in that @xmath743 combining and with , we get @xmath744 now in lemma [ lem : rij ]
concludes the proof .
[ lem : rij22 ] for any integers @xmath683 , @xmath745 for a fixed partition @xmath746 let @xmath747 with @xmath748 and @xmath749 be the number of blocks @xmath750 such that @xmath751 and @xmath752 .
we define @xmath753 for @xmath754 and @xmath755 and @xmath756 in the same way . for a possible combination of fixed numbers @xmath757 the number of partitions @xmath711
having this form is less than @xmath758 to get this inequality , we have used the fact that @xmath759 whence one of the factors in the denominator is at least @xmath760 . for a similar reason
there must be a factor in the denominator that is at least @xmath761 .
moreover , there are less than @xmath762 possible choices for @xmath757 , which concludes the proof .
note that we have not used the first and the third condition of the definition of @xmath699 in the proof of lemma [ lem : rij22 ] , whence the inequality even holds for a larger class of partitions .
now we are prepared for the proofs of theorem [ thm : abstract ] and corollary [ corol : univariat ] .
we aim at applying proposition [ prop : abstractbound ] . combining lemma
[ lem : rij22 ] and assumption , we get @xmath763 a straightforward computation and the inequality @xmath764 for @xmath765 show that @xmath766 where the right - hand side converges .
thus , theorem [ thm : abstract ] is a consequence of proposition [ prop : abstractbound ] . from ( * ? ?
* theorem 3.1 ) we know that @xmath767 where we can bound the right - hand side in the same way as in the proofs of proposition [ prop : abstractbound ] and theorem [ thm : abstract ] .
in the following we use the general normal approximation results of the previous section to derive central limit theorems for geometric functionals of the boolean model . we establish central limit theorems under the minimal moment assumption , but we need a stronger moment assumption in order to derive rates of convergence . for the berry - esseen bounds we assume that the typical grain @xmath44 of the boolean model satisfies the moment assumption @xmath768 for a fixed @xmath769 $ ] .
this allows us to state central limit theorems with rates of convergence depending on @xmath770 .
[ thm : multigeneral ] let @xmath771 be geometric functionals on @xmath10 and let @xmath772 .
assume that is satisfied and let @xmath723 be an @xmath26-dimensional centred gaussian random vector with covariance matrix @xmath773 given by .
then @xmath774 if is satisfied , there is a constant @xmath775 depending on @xmath776 and @xmath770 such that @xmath777 for @xmath70 with @xmath109 .
[ remboundsi]we will see in the proof of theorem [ thm : multigeneral ] that the translation invariance of @xmath771 is only used to ensure the existence of an asymptotic covariance matrix .
hence , such a multivariate central limit theorem still holds for functionals @xmath771 which are not translation invariant , if we can establish the existence of an asymptotic covariance matrix . in this case
the rate of convergence depends on the rate of convergence for the covariances . in the univariate case we can rescale by the square root of the variance , whence the existence of the asymptotic variance is not necessary .
thus , translation invariance of the functional is not required .
we only need to assume that the variance does not degenerate as @xmath97 , which , for instance , holds under the conditions of section [ secpos ] .
[ thm : univariat ] assume that is satisfied and let @xmath4 be an additive , locally bounded , and measurable functional on @xmath10 with constants @xmath778 and @xmath779 such that @xmath780 for @xmath70 with @xmath781 .
denote by @xmath723 a standard gaussian random variable .
then @xmath782 if is satisfied , there is a constant @xmath783 depending on @xmath784 and @xmath770 such that @xmath785 for @xmath70 with @xmath786 .
[ remboundsii]by replacing in the volume of @xmath3 by the volume of its inball , we obtain a rate of order @xmath787 . comparing and
, we see that for @xmath788 and @xmath789 the rate of convergence in the multivariate case is weaker than in the univariate case .
this is caused by the slow rate of convergence in theorem [ thm : variance ] since we need to bound @xmath790 in order to apply theorem [ thm : abstract ] . in the univariate analogue , which is corollary [ corol : univariat ] , we normalize with the exact variance and
do not have such a term .
if we replace the gaussian random vector @xmath723 by a centred gaussian random vector @xmath791 , having the covariance matrix of @xmath792 , the sum above vanishes and we obtain @xmath793 which is the same rate as in the univariate case . for @xmath794
we obtain by choosing @xmath795 as test function in the definition of the @xmath671-distance that @xmath796 hence proposition [ prop : optimalrate ] shows that the rate in is optimal for @xmath788 and @xmath797 .
we organize the proofs of theorem [ thm : multigeneral ] and theorem [ thm : univariat ] such that we first impose the moment assumption and establish and . in a second step
we prove that convergence in distribution is still obtained ( without convergence rates ) under the weaker moment assumption .
from now on , we write @xmath798 for @xmath799 , @xmath800 , and @xmath801 .
it is a direct consequence of that @xmath802 is the @xmath803-th kernel of the wiener - it chaos expansion of the poisson functional @xmath804 .
the assumption @xmath805 and the integrability condition are satisfied since the kernels are bounded by for every @xmath70 and the measure of the grains hitting @xmath3 is also finite . in the sequel we check assumption for the cases @xmath806 and @xmath807 separately .
we start with the first case .
let @xmath794 and @xmath806 . from in lemma [ lem : boundintrinsic ]
it follows that @xmath808 with non - empty sets @xmath809 , @xmath810 , depending on @xmath29 .
every @xmath811 is contained in at least two of these sets . by removing the index @xmath50 from the sets until it occurs only in one set , we increase the integral and can use lemma [ lem : kinematic ] to integrate over @xmath812 . due to the special structure of @xmath806
, we obtain by iterating this step and using the abbreviation @xmath813 that @xmath814 for a fixed @xmath815 , lemma [ lem : translativeintegral ] implies the second inequality in @xmath816\\ & \leq ( d+1)\gamma \beta_1 \e\left [ \bigg(\sum_{r=0}^d v_r(z_0)\bigg)^2\right ] \sum_{s=0}^d v_s(k_2\cap w ) .
\end{split}\ ] ] putting @xmath817 and applying lemma [ lem : translativeintegral ] again , we get @xmath818\\ & \leq ( d+1)\gamma \beta_1 c_9 ^ 2\ \e\left [ \bigg ( \sum_{r=0}^d v_r(z_0 ) \bigg)^2\right ] \sum_{s=0}^d v_s(w ) = c_{10 } \sum_{r=0}^d v_r(w)\end{aligned}\ ] ] with a constant @xmath819 depending on @xmath233 . finally , since @xmath820 for @xmath746 we have @xmath821 with @xmath822 and @xmath823 .
now it follows from lemma [ lem : boundinradius ] that there is a constant @xmath824 depending on @xmath771 and @xmath233 such that @xmath825 for @xmath70 with @xmath109 . for @xmath826
we obtain from in lemma [ lem : boundintrinsic ] and lemma [ lem : kinematic ] as above that @xmath827 a further application of lemma [ lem : translativeintegral ] yields the second inequality in @xmath828\\ & \qquad \leq ( d+1)\gamma \beta_1 { \mathbb{e}}\bigg[\bigg(\sum_{r=0}^d \min\{v_r(z_0),v_r(w)\}\bigg)^3 \sum_{s=0}^d v_s(z_0)\bigg ] \sum_{u=0}^d v_u(w).\end{aligned}\ ] ] consequently , we have @xmath829 with @xmath830 \sum_{u=0}^d v_u(w)\ ] ] and @xmath831 .
since @xmath832 \sum_{u=0}^d v_u(w)\\ & \leq { \mathbb{e}}\bigg[\bigg(\sum_{r=0}^d v_r(z_0)\bigg)^3 \sum_{s=0}^d \frac{\min\{v_s(z_0),v_s(w)\}}{v_d(w)}\bigg ] \sum_{u=0}^d \frac{v_u(w)}{v_d(w)}\\ & \leq { \mathbb{e}}\bigg[\bigg(\sum_{r=0}^d v_r(z_0)\bigg)^3 \sum_{s=0}^d \frac{v_s(z_0)^{\varepsilon } v_s(w)^{1-\varepsilon}\}}{v_d(w)}\bigg ] \sum_{u=0}^d \frac{v_u(w)}{v_d(w)},\end{aligned}\ ] ] lemma [ lem : boundinradius ] and the moment assumption imply that there is a constant @xmath833 depending on @xmath771 , @xmath233 , and @xmath770 such that @xmath834 for @xmath70 with @xmath835 . if we rescale the poisson functionals by @xmath836 , and together with and imply that holds with @xmath837 and @xmath838
. then @xmath839 is of the order @xmath840 .
now is a consequence of theorem [ thm : variance ] and theorem [ thm : abstract ] . by the same arguments as in the previous proof and analogous choices for @xmath841 , the conditions of corollary [ corol : univariat ] are satisfied with @xmath842 .
it follows from assumption that @xmath843 for @xmath70 with @xmath781 . combining this with and concludes the proof . for @xmath70
we define the set @xmath844 and the stationary boolean model @xmath845 the idea of the proofs of theorem [ thm : multigeneral ] and theorem [ thm : univariat ] under the weaker moment assumption is to approximate the boolean model @xmath0 by the boolean model @xmath846 .
a similar truncation has been used in @xcite to prove the central limit theorem for the volume of a more general boolean model based on a poisson process of cylinders .
we prepare the proofs by the following lemma . [
lem : approximation ] let @xmath4 be an additive , locally bounded , and measurable functional on @xmath10 and assume that is satisfied .
then @xmath847 and @xmath848 define @xmath849 further we define @xmath850 for @xmath851 and @xmath852 if there is a @xmath853 such that @xmath854 . in view of lemma [ lem : kernels ] and the fock space representation the assertions of this lemma are equivalent to @xmath855 and @xmath856 which we shall prove in the following . for @xmath133
we have @xmath857 our aim is to apply the dominated convergence theorem to the outer integral . for any @xmath858 it follows from lemma [ lem : boundintrinsic ] , lemma [ lem : translativeintegral ] , and lemma [ lem : kinematic ] similarly as in that @xmath859 where we have used that for the boolean model @xmath846 we can take the same constants as for @xmath0 .
the right - hand side of the previous inequality is uniformly bounded for @xmath109 because of lemma [ lem : boundinradius ] and it is integrable due to .
thus limit and summation in can be interchanged . by the same arguments , the second inequality above yields .
next we show that @xmath860 for any @xmath858 .
for @xmath861 we have @xmath862.\end{aligned}\ ] ] let us denote by @xmath863 the grains of @xmath1 that intersect @xmath864 and are not in @xmath865 .
then @xmath866 is a poisson distributed random variable with mean @xmath867 since @xmath868 , it follows from the inclusion - exclusion formula that @xmath869 denoting by @xmath870 the distribution of the restriction of @xmath1 to @xmath865 , we obtain by and the monotonicity of the intrinsic volumes that @xmath871 applying and the monotonicity of the intrinsic volumes yields @xmath872 since the restrictions of @xmath1 to @xmath865 and to its complement are stochastically independent , altogether we have that @xmath873 \sum_{i=0}^d v_i(k_1\cap\hdots\cap k_n\cap w)\\ & \leq \frac{\hat\beta(\psi)}{n!\sqrt{v_d(w ) } } \left(\exp(p_w(k_1))-1\right ) \sum_{i=0}^d v_i(k_1\cap\hdots\cap k_n\cap w)\end{aligned}\ ] ] with @xmath874 and @xmath875 . if there is a @xmath853 such that @xmath854 , we have @xmath876 , and it follows from lemma [ lem : boundintrinsic ] that @xmath877 for a fixed @xmath858 and @xmath96 sufficiently large such that @xmath878 , we have that @xmath879 lemma [ lem : translativeintegral ] and lemma [ lem : kinematic ] imply that @xmath880 then lemma [ lem : boundinradius ] and @xmath881 as @xmath97 show that the right - hand side vanishes for @xmath97 .
this proves so that the dominated convergence theorem yields , which concludes the proof . the triangle inequality for the @xmath671-distance yields @xmath882 in the sequel we show that both terms on the right - hand side of vanish as @xmath97 .
by ( * ? ? ?
* lemma 5.5 ) the first expression is bounded by @xmath883 where @xmath884 stands for the euclidean norm in @xmath885 . since , by lemma [ lem : approximation ] , the first factor is bounded and the second factor vanishes as @xmath97 , the first expression on the right - hand side of vanishes as @xmath97 . the restriction of @xmath1 to @xmath865 is a poisson process in @xmath886 of intensity @xmath35 and with the typical grain @xmath887 given by @xmath888 obviously , @xmath887 satisfies . by applying theorem [ thm : abstract ] to the vector @xmath889 of poisson functionals depending on the restriction of @xmath1 to @xmath865 , we shall prove that @xmath890 theorem [ thm : abstract ] yields this without a rate of convergence if @xmath891 for @xmath794 and if holds with a fixed @xmath718 and @xmath892 depending on @xmath3 such that @xmath839 tends to zero as @xmath97 .
the condition is satisfied because of lemma [ lem : approximation ] and theorem [ thm : variance ] .
the inequalities and also hold for the boolean model @xmath846 with the same @xmath893 as in the proof of under assumption and @xmath894\sum_{u=0}^d v_u(w)\ ] ] with @xmath895 .
this is the case since the derivations of and require only finite second moments and for the boolean model @xmath846 all inequalities proven in section [ sec:3 ] hold with the same constants as for the original boolean model @xmath0 .
consequently , is satisfied with @xmath896 and @xmath838 .
since only requires that the second moments are finite , we obtain that @xmath897 tends to zero as @xmath97 . on the other hand
, @xmath898 is equivalent to @xmath899=0.\ ] ] this expectation can be rewritten as @xmath900 for @xmath901 we have @xmath902 , for @xmath903 , and therefore @xmath904 which is independent of @xmath3 and integrable with respect to @xmath39 . for any fixed @xmath59
the left - hand side vanishes as @xmath97 so that the dominated convergence theorem implies and hence @xmath839 tends to zero as @xmath97 .
finally , theorem [ thm : abstract ] yields , which concludes the proof of theorem [ thm : multigeneral ] . as discussed in remark [ remboundsii ] , still holds if we replace the centred gaussian random vector @xmath723 with the asymptotic covariance matrix by a centred gaussian random vector @xmath791 with the exact covariance matrix .
this can be even done if the functionals are not translation invariant since in this case we do not need theorem [ thm : variance ] .
the second part of the proof of theorem [ thm : multigeneral ] still holds because does not apply in this situation .
this means that under condition for additive , locally bounded , and measurable functionals @xmath771 , @xmath905 for @xmath906 and a centred gaussian random variable @xmath791 with variance @xmath907 the previous remark implies that @xmath908 it follows from the definition of the @xmath671-distance that for random vectors @xmath663 and any @xmath909 ,
@xmath910 with @xmath911 and a standard gaussian random variable @xmath723 , this yields @xmath912 since @xmath913 is bounded by assumption , concludes the proof . in theorem [ thm : univariat ]
it is possible to weaken the assumption that the poisson process is stationary . in the proof
we only need to find upper bounds for the kernels and some integrals .
this is , for instance , still possible if the intensity measure is of the form @xmath914 with a non - negative bounded function @xmath915 .
now we always get upper bounds if we replace this intensity measure by the measure in with @xmath916 . for the multivariate central limit this argument does not work in general since its proof makes use of theorem [ thm : variance ] , which depends on the translation invariance of the intensity measure .
but if one can prove by other methods the existence of an asymptotic covariance matrix , it is still possible to weaken the stationarity assumption as described above .
lemma [ ldecomp ] is implied by lemma [ geomform ] below .
first , we provide two auxiliary results .
[ zeroexception ] let @xmath917 , @xmath918 , be convex bodies with non - empty interiors .
then , for @xmath919-almost all @xmath920 , if @xmath921 , then @xmath922
. the assertion is proved by induction over @xmath918 . for @xmath923 ,
the assertion holds , since any @xmath924 such that @xmath925 and @xmath926 is contained in the boundary of @xmath927 , which has @xmath37-dimensional hausdorff measure zero .
we assume that @xmath928 has been proved for some @xmath918 and @xmath929 are given .
let us denote by @xmath930 the set of all @xmath931 such that @xmath928 is satisfied for @xmath932 .
further , let @xmath933 be the set of all @xmath934 such that @xmath935 is not satisfied for @xmath929
. then @xmath936 let @xmath937 , @xmath938 and set @xmath939 . then @xmath940 and @xmath941 .
in particular , @xmath942 and @xmath943 , since @xmath938 .
but then @xmath944 , which implies , for fixed @xmath945 , that @xmath946 lies in a subset of @xmath2 of @xmath947-measure zero .
therefore the inner integral in vanishes , which proves @xmath948 .
for the following lemma , we use basic notions from geometric measure theory . [ rectify ] let @xmath949 , let @xmath932 be convex bodies in @xmath2 . if @xmath950 , then for @xmath919-almost all translates @xmath951 , the intersection @xmath952 has finite @xmath953-dimensional hausdorff measure . for @xmath954 ,
the intersection is the empty set for almost all translation vectors .
since for @xmath906 there is nothing to show , we assume that @xmath955 .
let @xmath956 , let @xmath957 be the compact image set of the lipschitz map @xmath958 , @xmath959 .
then the assumptions of the coarea theorem ( * ? ? ?
* theorem 3.2.22 ( 2 ) ) are satisfied .
thus , for @xmath919-almost all @xmath960 , the set @xmath961 has finite @xmath962 measure .
identify @xmath963 with @xmath964 and denote by @xmath965 the projection to the first component , which is a lipschitz map .
then @xmath966 has finite @xmath953-dimensional hausdorff measure for @xmath919-almost all @xmath960 .
( if @xmath967 , the intersection is the empty set . )
the assertion for @xmath954 easily follows from the one for @xmath968 .
recall that @xmath455=\{1,\ldots , n\}$ ] for @xmath133 .
[ geomform ] let @xmath969 be convex bodies with non - empty interiors , @xmath457 , and let @xmath970 .
then @xmath971\\ |i|\le d - i } } { \phi_i}\left(\bigcap_{r\in i}k_r;\cdot\cap \bigcap_{r\in i}\partial k_r\cap\bigcap_{s\notin i}k_s^\circ\right),\ ] ] for almost all translates @xmath376 of @xmath972 for @xmath973 .
there is nothing to show for @xmath428 , thence we assume that @xmath116 . by lemma
[ zeroexception ] and lemma [ rectify ] , we can assume that @xmath125 have a common interior point and for @xmath974 $ ] each intersection @xmath975 has finite @xmath976-dimensional hausdorff measure if @xmath977 , and otherwise is the empty set .
since @xmath978 is concentrated on the boundary of @xmath979 , the measure property yields that @xmath980 } { \phi_i}\left(k_1\cap\ldots\cap k_n;\cdot\cap \bigcap_{r\in i}\partial k_r\cap\bigcap_{s\notin i}k_s^\circ\right).\ ] ] the intersection @xmath981 is open , @xmath982 , and @xmath983 .
hence , since @xmath984 is locally defined ( see @xcite ) , it follows that @xmath980 } { \phi_i}\left ( \bigcap_{r\in i } k_r ; \cdot\cap\bigcap_{r\in i}\partial k_r\cap \bigcap_{s\notin i}k_s^\circ\right).\ ] ] since @xmath985 has finite @xmath976-dimensional hausdorff measure for @xmath986 , and is the empty set for @xmath987 , we conclude that if @xmath988 , then @xmath985 has @xmath803-dimensional hausdorff measure zero . a special case of ( * ? ? ?
* theorem 5.5 ) then yields that @xmath989 which completes the proof .
in section [ sintegra ] , we introduced certain measures in , , and whose definitions involve an integration with respect to the intensity measure of the underlying stationary poisson process .
we now provide an alternate description ( without detailed proofs ) of these measures in terms of mixed measures of translative integral geometry . throughout this section
, we assume that condition is satisfied . for @xmath422 , @xmath990 , and @xmath991 with @xmath992
we denote by @xmath993 the mixed measures of translative integral geometry .
subsequently , we shall repeatedly use the translative integral formula @xmath994 which holds for an arbitrary non - negative , measurable function @xmath628 ( see ( * ? ? ?
* ( 6.16 ) ) ) .
in addition to the properties of mixed measures stated in ( * ? ? ?
* theorem 6.4.1 ( b ) ) , we need the fact that the mixed measures are translation covariant ( independently in each argument ) , which follows from ( * ? ? ? * ( 6.14 ) ) together with an approximation argument .
moreover , we use the special form of the intensity measure without further mentioning . applying and the decomposability property of the mixed measures ( of translative integral geometry ) , stated in (
* theorem 6.4.1 ) , the range of summation is effectively restricted to @xmath995 and we obtain @xmath996 here and in the following , we denote by @xmath997 independent copies of the typical grain . now let @xmath429 and @xmath430 . then , arguing similarly , we can rewrite @xmath431 in the form @xmath998 in particular , for @xmath433 we thus recover as a very special case .
next we turn to the measure @xmath437 introduced in , for @xmath434 , @xmath430 , @xmath435 , and @xmath436 .
if @xmath999 , we obtain @xmath1000 in the case @xmath1001 , we get @xmath1002 for special choices of indices , these formulas simplify .
for instance , is recovered from the special case @xmath1003 and @xmath999 .
now we use that @xmath1004 ( see ( * ? ? ?
* p 229 ) ) .
if @xmath1005 , we obtain @xmath1006 in the particular case @xmath1007 , this yields @xmath1008 if also @xmath1009 , we recover . the special case @xmath1010 can be discussed similarly or
can be recovered by symmetry considerations .
it appears to be desirable to express the measures @xmath412 in terms of geometric mean values associated with a ( single ) typical grain .
we have seen that this is possible for @xmath1011 , for @xmath1012 ( see ) , for @xmath1013 ( see ) , and , in the isotropic case , for @xmath1014 , @xmath422 ; see , , and .
we do not know whether this can be achieved also for some of the other measures .
in section [ sintegra ] it was shown that @xmath1015 is close to @xmath405 if @xmath96 is large , where @xmath405 and @xmath406 are given in and by their series expansions . throughout this section ,
we assume that is satisfied .
theorem [ tdeltaij ] provides a succinct description of @xmath405 in terms of the finite measures @xmath412 . for a stationary and isotropic boolean model
, the local quantities @xmath406 are a crucial ingredient in the local covariances provided in theorem [ tsigmaijw ] . therefore , we are interested in describing the deviations @xmath1016 in more detail . recall that @xmath1017 , @xmath1018 , is the _ set covariance _ ( function ) of @xmath1019 .
clearly , we have @xmath1020 and @xmath1021 as @xmath97 , for @xmath1022 .
since the measures @xmath412 are bounded and @xmath1023 , for all @xmath43 , it follows from theorem [ tdeltaij ] that @xmath1024 as @xmath97 .
we also point out that the subsequent arguments show that @xmath1025 .
now we start to analyze the remainder terms @xmath1026 in more carefully . by definition
we have @xmath1027 where @xmath1028 in the simple case where @xmath467 and @xmath451 , we can use @xmath1029 to obtain from that @xmath1030 and therefore @xmath1031 . in the case
@xmath465 we obtain by a similar argument as in the proof of theorem [ tdeltaij ] that @xmath1032 and hence we get @xmath1033 .
next we consider the case where @xmath467 and @xmath1034 .
define @xmath1035 and @xmath455:=\{0,1,\ldots , n\}$ ] .
lemma [ ldecomp ] implies that @xmath1036\\ |j|=l } } \iint \i\{z\in \cap_{r\notin j}k^\circ_r\}\ , \,\phi_j(k_j;dz)\,\lambda^n(d(k_1,\ldots , k_n)),\end{aligned}\ ] ] where @xmath1037 for @xmath475 .
we split the summation into two parts , @xmath1038 where the first summand takes into account all subsets @xmath123 with @xmath1039 and the second summand those subsets with @xmath1040 . by symmetry ,
the first summand is given by @xmath1041 carrying out the integration with respect to @xmath1042 and using the form of @xmath233 , we obtain that @xmath1043 using and then the invariance of @xmath233 under translations , we get @xmath1044 therefore @xmath1045 in the same way we obtain that @xmath1046 this finally shows that , for @xmath422 , @xmath1047 next we note that , for @xmath429 , @xmath1048 we do not prove this fact , but proceed directly to summarize the similar but more involved case where @xmath434 . for the moment we fix @xmath133 .
lemma [ ldecomp ] implies that @xmath1049\\ |i|=k}}\sum_{\substack{j\subset [ n]\\ |j|=l } } & \iiint \i\{y\in \cap_{r\notin i}k^\circ_r , z\in \cap_{r\notin j}k^\circ_r\}\\ & \phi_i(k_i;dy)\,\phi_j(k_j;dz)\,\lambda^n(d(k_1,\ldots , k_n)).\end{aligned}\ ] ] distinguishing the cases @xmath1050 , @xmath1051 , @xmath1039 and @xmath1040 , we split the summation into four parts , @xmath1052 where the first and the fourth summand are for the cases @xmath1053 and @xmath1054 , and the second and the third summand refer to the symmetric cases @xmath1055 and @xmath1056 so that the third summand does not require further consideration .
we define [ tremainder ] let @xmath0 be a stationary boolean model such that and are satisfied
. let @xmath1019 .
then the remainder terms are given by @xmath1061 if @xmath1062 and @xmath451 , by for @xmath467 and @xmath422 , by for @xmath429 and @xmath451 , and by @xmath1063 here @xmath1064 and @xmath1065 are given by and , respectively .
together with theorem [ tsigmaijw ] the preceding theorem yields explicit integral representations for @xmath1066 for an isotropic boolean model .
we restrict ourselves to giving the formulas for the remainder terms in the case @xmath524 . in higher dimensions
the formulas get more complicated .
[ tremainderd=2 ] let @xmath0 be a stationary boolean model such that and are satisfied .
assume that @xmath524 , and let @xmath1067 .
then @xmath1068 .
moreover , @xmath1069 if the typical grain is also isotropic , then @xmath1070 | let @xmath0 be a boolean model based on a stationary poisson process @xmath1 of compact , convex particles in euclidean space @xmath2 .
let @xmath3 denote a compact , convex observation window .
for a large class of functionals @xmath4 , formulas for mean values of @xmath5 are available in the literature .
the first aim of the present work is to study the asymptotic covariances of general geometric ( additive , translation invariant , and locally bounded ) functionals of @xmath6 for increasing observation window @xmath3 , including convergence rates .
our approach is based on the fock space representation associated with @xmath1 . for the important special case of intrinsic volumes ,
the asymptotic covariance matrix is shown to be positive definite and can be explicitly expressed in terms of suitable moments of ( local ) curvature measures in the isotropic case .
the second aim of the paper is to prove multivariate central limit theorems including berry - esseen bounds .
these are based on a general normal approximation result obtained by the malliavin - stein method .
* key words : * boolean model , intrinsic volume , additive functional , curvature measure , covariance matrix , central limit theorem , fock space representation , malliavin calculus , wiener - it chaos expansion , berry - esseen bound , integral geometry * msc ( 2010 ) : * primary : 60d05 ; secondary : 60f05 , 60g55 , 60h07 , 52a22 |
the discoveries of the first @xmath4-ray afterglow ( costa et al . 1997 ) and optical transient ( ot ) ( van paradijs et al . 1997 ) of a gamma - ray burst ( grb ) have led to a major breakthrough in grb research .
the determination of a redshift of 0.835 for ( metzger et al .
1997 ) , and the subsequent determination of redshifts of more than a dozen bursts with a median redshift of @xmath51.0 , have firmly established their cosmological origin ( e.g. kulkarni et al . 2000 and references therein ) . in several cases supernovae ( sne )
have been connected to grbs .
attention to this connection was , from the observational side , first drawn in the remarkable case of that was associated with ( galama et al .
furthermore , the light curves of grb 970228 and grb 980326 ( reichart 1999 , galama et al . 2000 ; castro - tirado and gorosabel 1999 ; bloom et al .
1999 ) have bumps that are consistent with a contribution from an underlying sn . in no cases
the presence of an underlying sn contribution to the light
curve has been firmly excluded .
the exclusion of a sn contribution from the light
curve of grb 990712 ( hjorth et al .
2000 ) was based on a wrong identification of the ot in the _ hst _ image .
a revised analysis indeed shows an underlying sn contribution to the light curve ( bjrnsson et al .
. moreover , for less than half ( @xmath6 30% ) of all grbs with well - determined coordinates have searches for optical counterparts been successful . in some cases
the lack of detection can be easily explained by too bright detection limits ( e.g. due to light from the moon , nearby bright stars or twilight ) , too slow reaction time or high foreground extinction in the galaxy .
however , grb 981226 ( frail et al .
1999 ) and grb 990506 ( taylor et al .
2000 ) displayed no optical afterglow down to r band limits of r=23 and r=23.5 less than half a day after the bursts ( jensen et al . 1999 ; pedersen et al .
1999 ) , and both had a detectable radio afterglow .
grb 970828 ( groot et al .
1998a ) had a weak x - ray counterpart , but no ot was detected down to an r - band limit of 23.8 only 3 hours after the burst .
it is still not known whether these ` dark ' bursts constitute the faint end of a ( continuous ) ot luminosity function or whether they represent a special class of grbs that are truly dark in the optical ( in other words a dichotomy similar to the radio - loud vs. radio quiet dichotomy for qsos ) . in this paper
we present the detection of the ot of grb 000630 .
grb 000630 was observed by ulysses , konus - wind , near , and the bepposax grbm june 30.02 ut ( hurley et al . 2000 ) .
grb 000630 had a duration of @xmath520 s , and a 25100 kev fluence of @xmath52@xmath710@xmath8 erg @xmath9 .
these gamma - ray properties are consistent with the long duration class of bursts ( hurley 1992 ; kouveliotou et al .
1993 ) . in sect .
[ observations ] and in sect .
[ results ] we present our optical observations and the results we have obtained .
we also discuss the implications of our results on our understanding of the @xmath6 70% of grbs with no detected optical counterparts .
finally we summarize our results in sect .
[ summary ] .
the ipn error - box of ( hurley et al .
2000 ) was observed in the r band with the 2.56-m nordic optical telescope ( not ) on 2000 june 30.90 ut ( 0.88 days after the burst ) using the andaluca faint object spectrograph ( alfosc ) . comparing with red and blue palomar optical sky survey ii exposures no ot
was found ( jensen et al .
further r band imaging of the field was carried out on july 1.9 ut .
one object was found to have faded about one magnitude from june 30.90 to july 1.9 .
subsequent deep r band imaging obtained on july 3.9 ut confirmed the transient nature of the object and hence established this object as the likely optical afterglow of grb 000630 ( jensen et al .
further deep r band imaging of the ot was obtained at the not on july 10.9 ut and on july 25.9 ut .
we have also obtained r band imaging with the u.s . naval observatory flagstaff station ( usnofs ) 1.0-m telescope and the calar alto ( ca ) 2.2-m telescope ( greiner et al . 2000 ) , and b and v images with the 3.5-m telescopio nazionale galileo ( tng ) .
the journal of observations is reported in table [ obslog ] .
.the journal of observations . [ cols="<,<,<,^,^,^,^ " , ] it is still not understood why ots have only been detected for about 30% of well localized grbs .
possible explanations are _
i ) _ there is a optically bright vs. optically dark dichotomy for grbs similar to the radio quiet vs. radio loud dichotomy for qsos ( i.e. a bimodal ot luminosity function ) , _ ii ) _ a large fraction of grbs occur at redshifts @xmath10 7 and are hence invisible in the optical due to ly-@xmath1 blanketing and absorption by intervening lyman - limit systems , _ iii ) _ a large fraction of grbs occur in highly obscured galaxies similar to the scuba - selected galaxies ( e.g. ivison et al .
2000 ) so that the optical emission never escapes the host galaxies , or _
iv ) _ the shape of the ot luminosity function is such that with the search strategies applied in the period 19972000 we would not expect to detect ots for more than 30% of the well localized grbs . in order to shed some light on this problem
we have compiled in table [ darkdata ] , mainly using the gcn circular archive[multiblock footnote omitted ] , the most stringent r band upper limits to magnitudes and their observation epochs relative to the grb epoch ( @xmath11 ) for 43 grbs with no detected ot .
the left panel in fig .
[ dark ] shows the upper limits plotted against @xmath11 for the 43 grbs .
also plotted is the light
curve of the ot of grb 000630 ( full - drawn line ) and its extrapolation to earlier times than the first observing epoch ( dashed line ) . as seen , except for grb 990316 , all upper limits are above the _ observed _ ( @xmath3 1 day after grb event ) light curve of grb 000630 .
however , the strong upper limit on the magnitude of grb 990316 is only valid for a 2@xmath72 arcmin@xmath12 region around a candidate lotis transient that may be unrelated to the grb .
the upper limits for grb 970815 , grb 970828 , grb 981226 , grb 990217 , grb 990506 , grb 990704 , grb 990806 , grb 991014 , grb 991105 , and grb 000615 are below the extrapolation of the light
curve to earlier times .
two of these , grb 991014 and grb 991105 , occurred in fields with relatively high foreground extinction ( with e(b@xmath2v ) values 0.268 and 0.598 respectively ) .
note here , that not all ot light curves peak at @xmath13t@xmath141 day ( guarnieri et al .
1997 , but see also galama et al .
2000 ; pedersen et al .
1998 ) . in the right panel of fig .
[ dark ] we show the histogram of r band magnitudes at @xmath13t=1 day ( in 0.5 mag bins ) of the 13 ots ( prior to and including grb 000630 ) that were detected earlier than 24 hr after the burst .
it is much more difficult to detect an ot at r=23 than at r=19.5 , and an ot having r=24 at @xmath13t=1 day would almost certainly not be detected with current search strategies .
therefore the observed distribution is _ not _ the real distribution of apparent ot magnitudes .
the real distribution must have a larger fraction of bursts in the fainter bins .
from the two plots in fig .
[ dark ] we conclude that the majority ( @xmath3 75% ) of grbs without detected ots are consistent with no detection if they were similar to dim bursts like grb 980329 , grb 980613 , and grb 000630 at optical wavelengths . in fig .
[ compare]a c we compare the properties of batse - detected grbs with and without detected ots . in fig .
[ compare]a we show the spectral hardness vs. duration plot for 1959 batse bursts ( shown as contours following jensen et al .
( 2001 ) ) . superimposed as filled circles and crosses are the positions of grbs with and without detected ots .
as seen , the two groups populate the same region in the long - duration part of the diagram . in fig .
[ compare]b we show the channel 3 fluence histogram in 0.1 dex bins for all batse bursts .
the grbs with and without detected ots again show similar distributions . finally , in fig .
[ compare]c the r band magnitude of the ot at @xmath11 = 1 day in 0.5 mag bins for grbs with detected ots or the upper limit for grbs without detected ots are plotted against the logarithm of the channel 3 fluence .
we have shifted the upper limits to @xmath11 = 1 day assuming a decay slope of @xmath15 .
intuitively one might expect a correlation between the energy released at gamma - ray wavelengths and in the optical _ at some level _ , although different redshifts and different amounts of dust obscuration will increase the scatter . however , fig .
[ compare]c does not indicate any correlation .
moreover , the upper limits for the majority of grbs with no detected ots do not place them in regions of the diagram that are not already populated by grbs with detected ots .
this means that based on the gamma - ray properties of the current sample we can not conclude that grbs with no detected ots belong to another class of grbs than grbs with detected ots .
the ot of grb 000630 was detected with the not 21.1 hours after the high energy event with an r band magnitude of [email protected] .
the ot subsequently followed a power - law decay characterized by a slope of @xmath1 = @[email protected] . from the deep r band image obtained 25.9 days after the event we find that there is no evidence for an underlying sn .
furthermore , there is no indication of a host galaxy down to a 5@xmath16 limit of r=25.2 right underneath the point source emission from the ot .
the closest galaxy is a blue galaxy located 2.0 arcsec north of the ot . for the problem of dark bursts
we find that the majority ( @xmath3 75% ) of grbs for which searches for optical afterglow have been unsuccessful are consistent with no detection if they were similar to bursts like grb 000630 at optical wavelengths .
we at present can not exclude nor confirm any of the explanations _ i
iv _ listed above ( see sect 3.5 ) for why some bursts are darker than others . in order to establish the reasons why most grbs searches for ots up till now have been unsuccessful , and thereby obtain a more complete understanding of optical afterglows to grbs and the relation between optical and high energy properties of grbs , it is therefore essential to conduct deep ( r@xmath17 24 at @xmath181.0 day ) follow - up imaging at optical wavelengths and also deep infra - red follow - up of a sample of well localized and homogeneously selected grbs .
this is feasible with the next generation of grb satellites and 8m class telescopes .
we thank our anonymous referee for comments that helped us improve the discussion of dark bursts significantly .
some of the data presented here have been taken using alfosc , which is owned by the instituto de astrofisica de andalucia ( iaa ) and operated at the nordic optical telescope under agreement between iaa and the nbifafg of the astronomical observatory of copenhagen .
support for this programme by the director of the nordic optical telescope , professor piirola , is much appreciated .
we acknowledge m. feroci for his assistance with ipn operations and f. frontera for his assistance with bepposax operations .
kh is grateful for ulysses support under jpl contract 958056 , for near support under nasa grant nag5 - 9503 , and for bepposax support under nasa grant nag5 - 9126 .
we thank f. hoyo for help in the observations at calar alto .
jg acknowledges the receipt of a marie curie research grant from the european commission .
this work was supported by the danish natural science research council ( snf ) .
99 andersen m.i . , hjorth j. , pedersen h. , et al . , 2000 , a&a 364 , l54 arnouts s. , dodorico s. , cristiani s. , et al . , 1999 ,
a&a 343 , 19 .
bjrnson g. , hjorth j. , jakobsson p. , christensen l. , holland s. , apjl submitted bloom j.s . , kulkarni s.r .
, djorgovski s.g . , et al . , 1999 ,
nature 401 , 453 castro - tirado , a. j. , & gorosabel , j. , 1999 , a&as , 138 , 449 costa e. , frontera f. , heise j. , et al .
, 1997 , nature 387 , 783 frail d.a . ,
kulkarni s.r .
, bloom j.s . , et al . , 1999 ,
apj 525 , l81 galama t.s .
, vreeswijk p.m. , paradijs j. van , et al .
1998 , nature 395 , 670 galama t.s .
, tanvir n. , vreeswijk p.m. , et al .
2000 , apj 536 , 185 gorosabel j. , 1999 , ph d. thesis , univ . of valencia .
gorosabel j. , castro - tirado a.j . ,
wolf c. , et al .
, 1998 , a&a 339 , 719 gorosabel j. , 2000 , priv .
greiner j. , henden a. , merlino s , et al .
, 2000 , gcn 743 groot p.j .
, galama t.j . , van paradijs j. , et al .
, 1998a , apj 493 , l27 groot p.j .
, galama t.j . , vreeswijk p.m. , et al . , 1998b , apj 502 , l123 guarnieri a. , bartolini c. , masetti n. , et al . , 1997 , a&a 328 , l13 henden a. , 2000 , gcn 742 hjorth j. , andersen m.i .
, pedersen h. , et al . , 1998 , gcn 109 hjorth j. , pedersen h. , jaunsen a.o .
, andersen m.i . , 1999 ,
a&as 138 , 461 hjorth j. , holland s. , courbin f. , 2000 , apjl 534 , 147 hurley k. , 1992 , aip conf .
265 , `` gamma - ray bursts '' , eds . w. paciesas and g. fishman , aip ( new york ) hurley k. , cline t. , mazets e. , et al .
, 2000 , gcn 736 ivison r.j .
, smail i. , barger a.j . , et al .
, 2000 , mnras 315 , 209 jensen b.l . , hjorth j. , pedersen h. , et al . , 1999 , gcn 190 jensen b.l . , pedersen h. , hjorth j. , et al
. , 2000a , gcn 739 jensen b.l . ,
fynbo j.p.u . , pedersen h. , al . , 2000b , gcn 747 jensen b.l . , fynbo j.p.u . , gorosabel j. , al . , 2001 , submitted to a&a klose s. , stecklum b. , masetti n. , et al
. , 2000 , apj in press kouveliotou c. , meegan c.a .
, fishman g.j .
, et al . , 1993 , apj 413 , l101 kulkarni s.r . , berger e. , bloom j.s . , et al .
, 2000a , to appear in proc . of the 5th huntsville grb symposium ( astro - ph/0002168 ) mszros p. , rees m.j .
, 1997 , apj 476 , 232 metzger m.r .
, djorgovski s.g .
, kulkarni s.r .
, et al . , 1997 ,
nature 387 , 878 mller p. , warren s.j .
, 1993 , a&a 270 , 43 palazzi r. , pian e. , masetti n. , 1998 , a&al 336 , 95 van paradijs j. , groot p.j . , galama t. , et al . , 1997 ,
nature 386 , 686 pedersen h. , jaunsen a.o . , grav t. , et al . , 1998 , apj 496 , 311 pedersen h. , hjorth j. , jensen b.l .
, jaunsen a.o . ,
holland s. , 1999 , gcn 352 reichart d.,e . , 1999 , apj 521 , l111 schlegel d.j .
, finkbeiner d.p . , davis m. , 1998 , apj 500 , 525 stetson p. , 1987
, pasp 99 , 191s stetson p. , 1997
, `` user s manual for daophot ii '' taylor g.b . ,
frail d.a . ,
kulkarni s.r . , et al . , 1998 ,
apjl 502 , 115 taylor g.b .
, bloom j.s . , frail d.a . , et al .
, 2000 , apj 537 , l17 yost s. , harrison f. , diercks a. , 2000 , gcn 748 | we present the discovery of the optical transient of the long duration gamma - ray burst grb 000630 .
the optical transient was detected with the nordic optical telescope 21.1 hours after the burst . at the time of discovery
the magnitude of the transient was r = [email protected] .
the transient displayed a power - law decline characterized by a decay slope of @xmath1 = @[email protected] . a deep image obtained 25 days after the burst shows no indication of a contribution from a supernova or a host galaxy at the position of the transient .
the closest detected galaxy is a [email protected] galaxy 2.0 arcsec north of the transient .
the magnitudes of the optical afterglows of grb 980329 , grb 980613 and grb 000630 were all r@xmath323 less than 24 hours from the burst epoch .
we discuss the implications of this for our understanding of grbs without detected optical transients .
we conclude that _ i ) _ based on the gamma - ray properties of the current sample we can not conclude that grbs with no detected ots belong to another class of grbs than grbs with detected ots and _ ii ) _ the majority ( @xmath3 75% ) of grbs for which searches for optical afterglow have been unsuccessful are consistent with no detection if they were similar to bursts like grb 000630 at optical wavelengths . |
the higgs boson was discovered at lhc in 2012 @xcite , and its mass and coupling constants turned out to be consistent with the predictions in the standard model ( sm ) @xcite .
however , although the higgs boson was found , the structure of the higgs sector remains unknown .
there are possibilities of non - minimal higgs sectors which contain additional higgs bosons .
they also can satisfy the current data .
in fact , in large number of new physics scenarios , extended higgs sectors are predicted .
the structure of the higgs sector is strongly connected to these new physics scenarios .
therefore , the higgs sector is a probe of new physics .
although parameters of an extended higgs sector are basically free , they can be constrained by imposing some compelling theoretical conditions , such as those of perturbative unitarity , vacuum stability and triviality .
the requirement of perturbative unitarity @xcite is known to give a conservative bound on the parameters of a model , beyond which the perturbation calculation does not work .
it goes without saying that the parameters can also be constrained by taking into account various experimental results such as electroweak precision data @xcite , observables in flavour physics @xcite , the higgs boson signal strength @xcite and so on . by the combination of the bounds from the theoretical consistency and those from the experimental data , a parameter space in extended higgs sectors
can be further restricted .
the theoretical constraints have been intensively studied in the two higgs doublet model ( thdm ) .
nice reviews for the thdms have been presented in refs .
previously , the unitarity bound was mainly studied for the model with a ( softly - broken ) discrete @xmath0 symmetry @xcite with cp - conservation @xcite .
the softly - broken discrete @xmath0 symmetry is phenomenologically important in order to avoid flavor changing neutral currents ( fcncs ) , under which there are four types of yukawa interactions @xcite .
bounds from vacuum stability and triviality have also been studied in refs .
@xcite and in refs .
@xcite , respectively , in @xmath0 symmetric thdms .
these theoretical bounds have been used to limit magnitudes of cross sections and decay rates of tree level processes @xcite , radiative corrections due to additional higgs bosons to the higgs couplings @xcite and one - loop induced processes @xcite .
however , analyses in the most general thdm without the @xmath0 symmetry is getting important as an effective description of more various new physics scenarios , such as supersymmetric sms with non - holomorphic yukawa couplings @xcite and also general models with cp - violation @xcite which is required for successful scenario of electroweak baryogenesis @xcite . in this letter , we investigate unitarity bounds in the most general thdm without the @xmath0 symmetry nor cp - conservation .
we calculate s - wave amplitudes for two - body elastic scatterings of nambu - goldstone ( ng ) bosons and physical scalar bosons at high energies for all possible initial and final states ( 14 neutral @xcite , 8 singly - charged @xcite and 3 doubly - charged states ) . by choosing appropriate bases of the scattering states which are deduced using the conserved quantum numbers at high energies such as the hypercharge , weak isospin and the third component of the latter @xcite
, the scattering matrix is given as a block - diagonalized form with at most @xmath1 submatrices .
thus , all of the eigenvalues of the scattering matrix can be easily evaluated numerically . the analytic result for the block diagonalized
s - wave matrix is consistent with that given in ref . @xcite . by requiring that each of the eigenvalues is not too large to break validity of perturbation calculation ,
the model parameters , e.g. , the masses of extra higgs bosons and mixing angles , can be constrained .
our results can be useful to constrain parameter spaces whenever one evaluates physics quantities in the most general thdm .
we then numerically demonstrate that the bound on the mass @xmath2 of the second lightest higgs boson , whatever it is , is obtained by inputting the mass of the discovered higgs boson @xmath3 under the assumption of non - zero deviations in the higgs boson coupling @xmath4 with the weak boson @xmath5 and @xmath6 from the sm value .
currently , the @xmath4 coupling was measured with @xmath7 accuracy by the lhc run - i experiment ( see ref .
@xcite and references therein ) , and that is expected to be measured more accurately at future collider experiments . if a deviation is detected in the @xmath4 coupling , we can obtain the constraint on the region of @xmath2 even without its direct discovery .
the most general higgs potential under the @xmath8 gauge symmetry is given by @xmath9 \notag\\ & + \left[\lambda_6|\phi_1|^2\phi_1^\dagger \phi_2+\lambda_7|\phi_2|^2\phi_1^\dagger\phi_2+\text{h.c.}\right ] , \label{pot_thdm1}\end{aligned}\ ] ] where @xmath10 ( @xmath11 ) are the isospin doublet scalar fields with hypercharge @xmath12 .
in general , @xmath13 , @xmath14 and @xmath15-@xmath16 are real , while @xmath17 and @xmath18-@xmath19 are complex .
thus , there are totally fourteen real parameters .
if there is a symmetry in the potential , the number of parameters is reduced .
for example , when the potential is exact ( softly - broken ) @xmath0 invariant under the transformation of @xmath20 , the @xmath17 , @xmath21 and @xmath19 ( @xmath21 and @xmath19 ) terms are forbidden . by using the @xmath22 invariance and rephasing the doublet fields , the vacuum expectation values ( vevs ) of the two doublet fields
can be taken to be real without loss of generality @xcite .
the two doublet fields are then described in terms of the component fields as @xmath23,\hspace{3mm}(i=1,2 ) , \label{eq : parametrizations}\end{aligned}\ ] ] where @xmath24 and @xmath25 are the vevs of @xmath26 and @xmath27 , respectively , which are satisfied @xmath28 gev . by introducing @xmath29 , two vevs
are described by @xmath30 and @xmath31 as the usual notation . in the following , both the vevs
are assumed to be non - zero , except for the case of the inert doublet model discussed in appendix .
the stationary conditions of the scalar potential are given as follows @xmath32 where the left hand side of the above equation for each @xmath33 is calculated by @xmath34 , \label{tadpole1}\\ % % & \left.\frac{\partial v}{\partial h_2}\right|_0 = vs_\beta^{}\left[m_2 ^ 2 -m^2\cos^2\beta + \frac{v^2}{2}\left(\lambda_2s^2_\beta + \lambda_{345}c^2_\beta+\lambda_6^rc^2_\beta\cot\beta + 3\lambda_7^rs_\beta^{}c_\beta^ { } \right)\right ] , \label{tadpole2}\\ % % & \left.\frac{\partial v}{\partial z_1}\right|_0 = -v\tan\beta\left.\frac{\partial v}{\partial z_2}\right|_0 = vs_\beta^{}\left[\text{im}\ , m_3 ^ 2 + \frac{v^2}{2}\left(\lambda_5^i s_\beta^{}c_\beta^ { } + \lambda_6^ic^2_\beta + \lambda_7^is^2_\beta \right)\right ] , \label{tadpole3}\end{aligned}\ ] ] where we used the abbreviation of @xmath35 and @xmath36 .
( [ tadpole1])-([tadpole3 ] ) , we introduced @xmath37 and @xmath38 from the stationary conditions in eq .
( [ stationaly ] ) , we can eliminate @xmath13 , @xmath14 and im@xmath17 in the higgs potential . in order to calculate the masses for the scalar bosons , it is convenient to introduce the so - called higgs basis @xcite defined as @xmath39 where @xmath40,\quad \psi=\left [ \begin{array}{c } h^+\\ \frac{1}{\sqrt{2}}(h_2'+ih_3 ' ) \end{array}\right ] , \label{higgs - basis}\end{aligned}\ ] ] where @xmath41 and @xmath42 are the ng bosons which are absorbed into the longitudinal components of @xmath43 and @xmath44 by the higgs mechanism , respectively .
the physical singly - charged scalar state is denoted as @xmath45 .
the three neutral states @xmath46 , @xmath47 and @xmath48 are not the mass eigenstates at this stage , which generally mix with each other .
we give the mass formulae for the higgs bosons in the most general case .
first , the squared mass of @xmath45 is calculated by @xmath49 next , the mass term for the neutral scalar states is expressed by the @xmath50 matrix as @xmath51 where each of the matrix elements is given by @xmath52 , \label{m11sq }
\\ m_{22}^2&=m^2+\frac{v^2}{4}\big [ ( \lambda_1+\lambda_2 - 2\lambda_{345})s^2_{2\beta } -2\lambda_6^r(s_{4\beta}^{}+\cot\beta ) -2\lambda_7^r(s_{4\beta}^{}-\tan\beta)\big ] , \label{m22sq}\\ m_{33}^2 & = m^2-v^2\left(\lambda_5^r + \frac{\lambda_6^r } { 2}\cot\beta + \frac{\lambda_7^r}{2}\tan\beta\right ) , \label{m33sq }
\\ m_{12}^2&=\frac{v^2}{2}\left[(\lambda_2 s^2_\beta-\lambda_1c^2_\beta + \lambda_{345}c_{2\beta}^ { } ) s_{2\beta}^ { } + 2\lambda^r_6(2c_{2\beta}^ { } -1)c^2_\beta + 2\lambda_7^r(2c_{2\beta}^ { } + 1)s^2_\beta \right ] , \label{m12sq } \\
% m_{13}^2 & = -\frac{v^2}{2}\left(\lambda_5^i s_{2\beta}^ { } + 2\lambda_6^i c^2_\beta + 2\lambda_7^i s^2_\beta \right ) , \\
m_{23}^2 & = -\frac{v^2}{2}\left(\lambda_5^ic_{2\beta}^ { } - \lambda_6^ic^2_\beta + \lambda_7^is^2_\beta \right ) .
\end{aligned}\ ] ] [ mateven ] the mass eigenvalues @xmath53
( @xmath54 - 3 ) are obtained by introducing the @xmath50 orthogonal matrix @xmath55 as @xmath56 where @xmath57 is assumed .
mass eigenstates for the neutral higgs bosons are also defined using @xmath55 as @xmath58 where @xmath59 is defined to be the sm - like higgs boson with the mass of about 125 gev , i.e. , @xmath60 gev . in the following ,
we represent @xmath59 and its mass @xmath61 by @xmath3 and @xmath62 , respectively . in the cp - conserving limit , @xmath63 and @xmath64 respectively
correspond to the additional cp - even ( @xmath65 ) and cp - odd ( @xmath66 ) higgs bosons .
the mass formulae for the cp - conserving case will be discussed in the next subsection .
we here note that @xmath55 can be described by three mixing angles @xcite . in our numerical analysis given in sec .
iv , the matrix elements of @xmath55 are derived by inputting the mass matrix elements given in eq .
( [ mateven ] ) . by using eqs .
( [ mchsq ] ) , ( [ m11sq ] ) , ( [ m22sq ] ) , ( [ m22sq ] ) , and ( [ m33sq ] ) , the five parameters ( @xmath67 and @xmath68 ) can be rewritten as follows @xmath69 therefore , we can choose thirteen input parameters as @xmath30 , @xmath31 , @xmath70 , @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath75 , @xmath76 and @xmath77 . instead of @xmath71 , @xmath72 and @xmath73 , we can take @xmath78 , @xmath79 and @xmath80 as inputs by @xmath81 in addition , we rewrite @xmath82 as @xmath83 in the cp - conserving limit ( @xmath84 and @xmath85 ) , @xmath86 , @xmath87 , @xmath88 and @xmath78 describe the physical squared masses for @xmath3 , @xmath65 and @xmath66 , and the mixing angle for the cp - even higgs state , respectively .
consequently , by using experimental values of @xmath30 ( @xmath89 gev ) and @xmath90 ( @xmath91 gev ) , eleven input parameters can be chosen as follows @xmath92 where @xmath93 are the complex phases of @xmath94 .
notice that @xmath95 is determined so as to keep @xmath96 gev .
we calculate the s - wave amplitude matrix for the elastic @xmath97 scatterings in the high energy limit , where the fields @xmath98 , @xmath99 , @xmath100 and @xmath101 represent either @xmath102 , @xmath103 , @xmath59 , @xmath63 , @xmath64 or @xmath45 . in this case , all the longitudinal components of the weak gauge boson states can be replaced by the corresponding ng boson states because of the equivalence theorem .
furthermore , only the scalar boson contact interactions contribute to the s - wave amplitude .
therefore , the calculation of the s - wave amplitude matrix is quite simply done just by extracting the coefficient of the scalar boson quartic terms .
there are 14 neutral , 8 singly - charged and 3 doubly - charged 2 body scalar boson channels .
first , the 14 neutral channels are expressed in the weak eigenbasis as @xmath104 where each submatrix is given by @xmath105 each of the submatrices is obtained in the following basis : @xmath106 where the indices @xmath107 and @xmath108 ( @xmath109 and @xmath110 ) run over 1 - 4 ( 1 - 3 ) .
each of the above bases give the submatrix indicated in the parenthesis .
we note that at the high energy limit , the hypercharge @xmath111 , the isospin @xmath112 and the third component of the isospin @xmath113 are used to classify the two body scattering states @xcite , namely @xmath114 with @xmath115 and @xmath116 ( @xmath117 ) with @xmath118 .
in fact , the above bases are obtained by finding the two scalar states which belong to the same set of the quantum numbers , i.e. , the states in the @xmath119 , @xmath120 , @xmath121 and @xmath122 bases respectively belong to the state with ( @xmath111 , @xmath112 , @xmath123 , @xmath124 , @xmath125 and @xmath125 .
next , eight ( positive ) singly - charged channels are expressed as : @xmath126 each of the submatrices and the eigenvalue @xmath127 are obtained in the following basis : @xmath128 where @xmath54 - 4 and @xmath129 - 3 .
similar to the discussion for the neutral states , the states in the bases @xmath130 , @xmath131 and @xmath132 respectively correspond to the states with ( @xmath111 , @xmath112 , @xmath133 , @xmath134 and @xmath135 .
finally , three ( positive ) doubly - charged channels are expressed as : @xmath136 we note that the negative charged states are obtained by taking charge conjugation for each positive charged channel , which give the same set of eigenvalues of the matrix for the corresponding positive charged channels .
when we impose symmetries in the potential , we obtain the block - diagonalized s - wave amplitude matrix with a smaller size of submatrices .
for example , if we assume the cp - invariance ( for the case of @xmath137 ) , the maximal size of the submatrices reduces into @xmath50 , or if we impose the ( softly - broken ) @xmath0 symmetry ( for the case of @xmath138 ) , the maximal size of the submatrices reduces into @xmath139 as they have already known in refs . @xcite . in order to constrain the parameters in the potential , we impose the following condition for each eigenvalue of the s - wave amplitude matrix : @xmath140 is conventionally taken to be 1/2 @xcite or 1 @xcite , and @xmath141 with all @xmath142 being real due to the hermitian nature of the s - wave amplitude matrix .
we call the bound given by the inequality ( [ lqtbound ] ) as the unitarity bound .
we note that the analytic expressions for the eigenvalues in eq .
( [ eigen ] ) are given by solving the fourth and third order equations .
in general , the solutions of such an equation is given as a too complicated form to explicitly show in this letter .
therefore , we do not show the explicit formula , and we numerically calculate the eigenvalues in the numerical study given in the next section .
we here discuss the constraint on the parameter space using the unitarity bound .
the unitarity bound in eq .
( [ lqtbound ] ) sets upper limits on @xmath142 ( @xmath143 ) given in eq .
( [ eigen ] ) , which are expressed by combinations of @xmath144 parameters .
using eqs .
( [ lam1])-([m12 ] ) , this constraint can be translated into the bound on physical parameters such as the masses of higgs bosons and mixing angles .
it is worthful to mention here that we can obtain an upper limit on the masses of extra higgs bosons even when the sm - like higgs boson @xmath145 coupling with the gauge bosons ( @xmath4 , @xmath146 ) slightly deviates from the sm prediction .
this can be intuitively understood as follows .
first , in the higgs basis defined in eq .
( [ higgs - basis ] ) , the kinetic terms for the higgs fields are given by @xmath147 where @xmath148 is the covariant derivative .
the higgs - gauge - gauge type vertex , i.e. , @xmath149 , only comes from the first term of eq .
( [ lkin ] ) at the tree level . in the mass eigenbasis ,
the ratio of the @xmath4 coupling @xmath150 in the thdm to that of sm is then expressed by using the rotation matrix @xmath55 given in eq .
( [ neut ] ) as @xmath151 we note that @xmath152 corresponds to @xmath153 in the cp - conserving case as it is seen in eq .
( [ sinba ] ) .
thus , the non - zero deviation in the @xmath4 couplings from the sm prediction comes from the mixing effect of neutral higgs bosons , i.e. , @xmath154 .
second , when there is no mixing among the neutral higgs bosons , the mass of @xmath3 and those of the extra higgs bosons @xmath63 and @xmath64 are schematically expressed as @xmath155 and @xmath156 , respectively , as it is seen in eqs .
( [ m11sq ] ) , ( [ m22sq ] ) and ( [ m33sq ] ) .
therefore , in the no mixing case , the upper bound on the masses of extra higgs bosons can not be obtained , because they can be taken to be as large as possible by using the @xmath70 dependence . in other words
, we can take the decoupling limit @xcite of the extra higgs bosons by taking @xmath157 .
on the other hand , if there is non - zero mixing among the neutral higgs bosons , i.e. , @xmath158 , @xmath70 dependence appears in @xmath159 which must be kept to be about ( 125 gev)@xmath160 .
therefore , we can not take a too large value of @xmath70 in that case , because we need a large cancellation of the @xmath70 contribution to @xmath159 by the @xmath155 term which must be excluded by the unitarity bound .
therefore , we can obtain an upper limit on the masses of extra higgs bosons as long as the @xmath4 coupling is deviated from the sm prediction . from the unitarity bound in the softly - broken @xmath0 symmetric case .
the left and right panels show the @xmath161 and @xmath31 dependences of the upper limit for fixed values of @xmath162 ( black ) , 3 ( blue ) , 5 ( green ) and 10 ( red ) and those of @xmath163 ( black ) , 0.97 ( blue ) and 0.95 ( red ) , respectively .
the solid ( dashed ) curve shows the case with @xmath164 ( 1 ) in eq .
( [ lqtbound ] ) . in both panels ,
we take @xmath165 , and scan the value of @xmath70 . ,
title="fig:",width=264 ] from the unitarity bound in the softly - broken @xmath0 symmetric case .
the left and right panels show the @xmath161 and @xmath31 dependences of the upper limit for fixed values of @xmath162 ( black ) , 3 ( blue ) , 5 ( green ) and 10 ( red ) and those of @xmath163 ( black ) , 0.97 ( blue ) and 0.95 ( red ) , respectively .
the solid ( dashed ) curve shows the case with @xmath164 ( 1 ) in eq .
( [ lqtbound ] ) . in both panels ,
we take @xmath165 , and scan the value of @xmath70 . , title="fig:",width=264 ] from the unitarity bound as a function of @xmath166 in the case without the cp - violating phases .
the black , blue and red curves show the case of @xmath163 , 0.97 and 0.95 , respectively .
we take @xmath165 and @xmath162 .
the left ( right ) panel shows the case of @xmath167 ( @xmath168 ) .
, title="fig:",width=264 ] from the unitarity bound as a function of @xmath166 in the case without the cp - violating phases .
the black , blue and red curves show the case of @xmath163 , 0.97 and 0.95 , respectively .
we take @xmath165 and @xmath162 .
the left ( right ) panel shows the case of @xmath167 ( @xmath168 ) .
, title="fig:",width=264 ] in the following , we numerically show the bound on the mass of the second lightest higgs boson denoted as @xmath2 by fixing @xmath169 gev and @xmath161 .
we assume that the sm - like higgs boson @xmath3 is the lightest of all the higgs bosons , so that @xmath2 is defined by @xmath170 we first consider the softly - broken @xmath0 symmetric and cp - conserving case , i.e. , @xmath171 and @xmath172 . in this case
, we have six free parameters @xmath173 , @xmath174 , @xmath175 , @xmath70 , @xmath31 and @xmath176 . in fig . [ uni1 ]
, we show the upper limit on @xmath2 as a function of @xmath161 ( left panel ) and @xmath31 ( right panel ) in the case of @xmath165 .
in the left ( right ) panel , each of the curves show the cases of @xmath31 ( @xmath161 ) to be 1 , 3 , 5 and 10 ( 0.99 , 0.97 and 0.95 ) .
the value of @xmath70 is scanned in the both figures .
the solid ( dashed ) curves show the case with @xmath164 ( 1 ) in eq .
( [ lqtbound ] ) for the comparison of the strength of the unitarity constraint with the two cases .
hereafter , we take @xmath164 . from these figures
, we can see that the stronger limit on @xmath2 is obtained when @xmath177 is taken to be larger values .
our results for this case are consistent with those given in ref .
@xcite . and @xmath2 plane in the case of @xmath178 ( upper - left ) , 0.5 ( upper - right ) , 1 ( lower - left ) and 2 ( lower - right ) .
for all the plots , we take @xmath179 , @xmath180 and @xmath162 , and scan the @xmath181 and @xmath182 parameters . the black ( circle ) , blue ( square ) and red ( triangle ) dots show the allowed points for the cases of @xmath183 , @xmath184 and @xmath185 , respectively .
, title="fig:",width=264 ] and @xmath2 plane in the case of @xmath178 ( upper - left ) , 0.5 ( upper - right ) , 1 ( lower - left ) and 2 ( lower - right ) . for all the plots ,
we take @xmath179 , @xmath180 and @xmath162 , and scan the @xmath181 and @xmath182 parameters .
the black ( circle ) , blue ( square ) and red ( triangle ) dots show the allowed points for the cases of @xmath183 , @xmath184 and @xmath185 , respectively .
, title="fig:",width=264 ] + and @xmath2 plane in the case of @xmath178 ( upper - left ) , 0.5 ( upper - right ) , 1 ( lower - left ) and 2 ( lower - right ) . for all the plots , we take @xmath179 , @xmath180 and @xmath162 , and scan the @xmath181 and @xmath182 parameters . the black ( circle ) , blue ( square ) and red ( triangle ) dots show the allowed points for the cases of @xmath183 , @xmath184 and @xmath185 , respectively . , title="fig:",width=264 ] and @xmath2 plane in the case of @xmath178 ( upper - left ) , 0.5 ( upper - right ) , 1 ( lower - left ) and 2 ( lower - right ) .
for all the plots , we take @xmath179 , @xmath180 and @xmath162 , and scan the @xmath181 and @xmath182 parameters .
the black ( circle ) , blue ( square ) and red ( triangle ) dots show the allowed points for the cases of @xmath183 , @xmath184 and @xmath185 , respectively .
, title="fig:",width=264 ] next , we consider the case without the softly - broken @xmath0 symmetry but with cp - conservation , i.e. , @xmath186 . in fig .
[ uni2 ] , we show the upper limit on @xmath2 as a function of @xmath187 in the case of @xmath188 and @xmath189 . the value of @xmath70 is taken to be @xmath190 ( left panel ) and @xmath191 ( right panel ) . in both the panels , the black ,
blue and red curves show the cases with @xmath163 , 0.97 and 0.95 , respectively .
clearly , the stronger bound is obtained in the case of @xmath191 ( right panel ) than the case of @xmath192 , where the appearance of kink at around @xmath193 , 0.8 and 1 is due to an interchange of eigenvalues which break the condition of eq .
( [ lqtbound ] ) .
when @xmath194 is taken to be larger than about 2.2 , there is no solution to satisfy the unitarity bound .
we confirm that the maximally allowed value of @xmath2 is obtained in the case of @xmath195 .
finally , we consider the most general case with cp - violation . in this case
, we have eleven input parameters with @xmath196 gev and @xmath96 gev as shown in ( [ input1 ] ) . in order to satisfy @xmath96 gev
, we scan the @xmath95 parameter for each fixed value of the other parameters . in fig .
[ uni3 ] , we show the allowed parameter regions from the unitarity bound on the @xmath197 and @xmath2 plane in the case of @xmath198 , @xmath199 and @xmath180 .
the upper - left , upper - right , lower - left and lower - right panels respectively show the cases with @xmath200 , 0.5 , 1 and 2 . in all the panels , the circle ( black ) ,
square ( blue ) and triangle ( red ) points satisfy the unitarity bound and @xmath183 , @xmath184 and @xmath185 , respectively .
the values of @xmath201 and @xmath182 are scanned .
we find that , in addition to the upper limit on @xmath2 , there is the lower limit especially in the case with @xmath202 .
the lower bound becomes higher when we take a larger value of @xmath203 .
the appearance of the lower limit can be understood in the following way .
if we take a non - zero value of @xmath204 , it gives a non - zero value of the off - diagonal mass matrix elements @xmath205 and @xmath206 ( see eq .
( [ mateven ] ) ) , which gives non - zero mixings and/or mass splittings among the neutral higgs bosons . on the other hand
, we now fix the value of @xmath161 , so that it restricts the possible amount of the mixing , and it also requires a non - zero mass splitting among the neutral higgs bosons . because the sm - like higgs boson @xmath3 which we suppose the lightest of all has the mass of 125 gev , the non - zero mass splitting turns out to be the lower limit on @xmath2
. we here comment on the constraint from electric dipole moments ( edms ) on the parameter space in the thdm . as it has been well known that if a model has an additional source of the cp - violation , its magnitude is constrained by edms . in refs .
@xcite , the constraint on parameter space from edms has been investigated in the softly - broken @xmath0 symmetric thdms . in the thdms ,
edms constrain the allowed region of @xmath204 depending on the parameter set .
recent study on the collider phenomenology of the cp - violating thdm is found in ref .
@xcite , in which the constraint on the complex parameter such as @xmath18 from edms and the unitarity bound turns out to be important .
we have investigated unitarity bounds in the most general two higgs doublet model without a discrete @xmath0 symmetry nor cp conservation . we have computed the s - wave amplitudes for two - body elastic scatterings of the ng bosons and physical higgs bosons at high energies for all possible initial and final states . by choosing the appropriate bases ,
the scattering amplitude matrix is given to be the block - diagonalized form , and thus the eigenvalues can be easily evaluated numerically .
we have constrained the parameter space of the model by using the unitarity bound . by fixing the mass of the discovered higgs boson @xmath3 to be 125 gev and assuming a non - zero deviation in the @xmath4 couplings from the sm values
, there is an upper limit on the mass @xmath2 of the second lightest higgs boson .
therefore , by using the precisely measured @xmath4 couplings at future collider experiments , we can constrain the allowed region of @xmath2 if a deviation of the @xmath4 coupling from the sm limit is found even without its direct discovery .
our results can be useful to constrain parameter spaces whenever one evaluates physics quantities in the most general thdm .
this work was supported , in part , by grant - in - aid for scientific research no .
23104006 ( sk ) and grand h2020-msca - rise-2014 no .
645722 ( non minimal higgs ) ( sk ) .
this work was also supported by a jsps postdoctoral fellowships for research abroad ( ky ) .
we present the mass formulae of the scalar bosons for specific cases of the thdm , i.e. , the case with cp - conservation but without the @xmath0 symmetry , and that with the unbroken @xmath0 symmetry usually referred as the inert doublet model .
when we take the limit of @xmath85 , the cp symmetry is restored in the higgs potential , and the potential is then described by the ten parameters .
the mass of @xmath45 is unchanged and is given in eq .
( [ mchsq ] ) . for the mass matrix of neutral higgs bosons , the @xmath50 matrix given in eq .
( [ massmat ] ) becomes a block - diagonalized form , i.e. , the @xmath139 plus @xmath207 form .
the first @xmath139 part with the basis of ( @xmath46,@xmath47 ) corresponds to the mass matrix for the cp - even higgs states , and the remained @xmath207 with the basis of @xmath208 does the squared mass of the cp - odd higgs boson . for the cp - even states , we obtain the mass eigenstates by introducing the mixing angle @xmath209 by @xmath210 where @xmath3 is defined as the sm - like higgs boson .
the squared masses of @xmath66 is given by @xmath211 masses for @xmath65 and @xmath3 and the mixing angle @xmath209 are respectively given by solving eqs .
( [ m11 ] ) , ( [ m22 ] ) and ( [ m12 ] ) with the replacement @xmath212 : @xmath213 consequently , eight input parameters can be chosen as follows @xmath214 and experimental values of @xmath30 and @xmath90 . in this parameter choice ,
the five parameters @xmath67 and @xmath215 are derived by eqs .
( [ lam1])-([m12 ] ) with the replacement @xmath212 . when we impose an exact @xmath0 symmetry in the potential , where the two doublets are transformed as @xmath217 and @xmath218 , the potential is given for the case with @xmath219 in eq .
( [ pot_thdm1 ] ) . in this case
, the phase of @xmath18 can be dropped without loss of generality , so that cp - violation does not occur .
the vev of @xmath27 must be assumed to be zero to avoid the spontaneous breakdown of the @xmath0 symmetry , otherwise the @xmath220 terms are generated through the @xmath144 terms , and the inert nature , i.e. , without couplings to sm particles , is then lost via the mixing with the active higgs field , namely , the sm - like higgs boson .
g. aad _ et al . _ [ atlas collaboration ] , phys .
b * 716 * , 1 ( 2012 ) .
s. chatrchyan
_ et al . _
[ cms collaboration ] , phys .
b * 716 * , 30 ( 2012 ) .
g. aad _ et al . _ [ atlas collaboration ] , phys .
d * 91 * , 012006 ( 2015 ) .
v. khachatryan _ et al . _
[ cms collaboration ] , eur .
j. c * 75 * , no . 5 , 212 ( 2015 ) . b. w. lee , c. quigg and h. b. thacker , phys .
d * 16 * , 1519 ( 1977 ) .
s. schael _ et al . _
[ aleph and delphi and l3 and opal and sld and lep electroweak working group and sld electroweak group and sld heavy flavour group collaborations ] , phys . rept . * 427 * , 257 ( 2006 ) .
y. amhis _ et al . _
[ heavy flavor averaging group ( hfag ) collaboration ] , arxiv:1412.7515 [ hep - ex ] . j. f. gunion , h. e. haber , g. l. kane and s. dawson , front .
phys . * 80 * , 1 ( 2000 ) .
v. d. barger , j. l. hewett and r. j. n. phillips , phys .
d * 41 * , 3421 ( 1990 ) .
y. grossman , nucl .
b * 426 * , 355 ( 1994 ) .
a. g. akeroyd , phys .
b * 377 * , 95 ( 1996 ) .
m. aoki , s. kanemura , k. tsumura and k. yagyu , phys .
d * 80 * , 015017 ( 2009 ) .
v. barger , h. e. logan and g. shaughnessy , phys .
d * 79 * , 115018 ( 2009 ) .
s. su and b. thomas , phys .
d * 79 * , 095014 ( 2009 ) .
n. g. deshpande and e. ma , phys .
d * 18 * , 2574 ( 1978 ) .
s. nie and m. sher , phys .
b * 449 * , 89 ( 1999 ) .
s. kanemura , t. kasai and y. okada , phys .
b * 471 * , 182 ( 1999 ) .
k. inoue , a. kakuto and y. nakano , prog . theor . phys . *
63 * , 234 ( 1980 ) .
n. chakrabarty , u. k. dey and b. mukhopadhyaya , jhep * 1412 * , 166 ( 2014 ) .
a. wahab el kaffas , p. osland and o. m. ogreid , phys .
d * 76 * , 095001 ( 2007 ) ; s. kanemura , s. moretti , y. mukai , r. santos and k. yagyu , phys . rev .
d * 79 * , 055017 ( 2009 ) ; c. y. chen and s. dawson , phys .
d * 87 * , 055016 ( 2013 ) ; c. w. chiang and k. yagyu , jhep * 1307 * , 160 ( 2013 ) ; d. lpez - val , t. plehn and m. rauch , jhep * 1310 * , 134 ( 2013 ) ; a. celis , v. ilisie and a. pich , jhep * 1312 * , 095 ( 2013 ) ; j. baglio , o. eberhardt , u. nierste and m. wiebusch , phys
. rev .
d * 90 * , no . 1
, 015008 ( 2014 ) ; p. m. ferreira , j. f. gunion , h. e. haber and r. santos , phys .
d * 89 * , no . 11 , 115003 ( 2014 ) ; p. m. ferreira , r. guedes , m. o. p. sampaio and r. santos , jhep * 1412 * , 067 ( 2014 ) ; l. basso , p. osland and g. m. pruna , jhep * 1506 * , 083 ( 2015 ) ; a. arhrib , r. benbrik , c. h. chen , m. gomez - bock and s. semlali , arxiv:1508.06490 [ hep - ph ] .
s. kanemura , k. tsumura , k. yagyu and h. yokoya , phys .
d * 90 * , 075001 ( 2014 ) .
s. kanemura , m. kikuchi and k. yagyu , phys .
b * 731 * , 27 ( 2014 ) ; s. kanemura , m. kikuchi and k. yagyu , nucl .
b * 896 * , 80 ( 2015 ) ; a. arhrib , r. benbrik , j. el falaki and a. jueid , arxiv:1507.03630 [ hep - ph ] . s. kanemura , y. okada , e. senaha and c .- p .
yuan , phys .
d * 70 * , 115002 ( 2004 ) .
s. kanemura , eur .
j. c * 17 * , 473 ( 2000 ) ; e. asakawa , o. brein and s. kanemura , phys .
d * 72 * , 055017 ( 2005 ) ; e. asakawa , d. harada , s. kanemura , y. okada and k. tsumura , phys .
d * 82 * , 115002 ( 2010 ) ; j. sola and d. lopez - val , nuovo cim .
c * 34s1 * , 57 ( 2011 ) ; a. arhrib , r. benbrik and n. gaur , phys . rev .
d * 85 * , 095021 ( 2012 ) ; b. swiezewska and m. krawczyk , phys . rev .
d * 88 * , no . 3 , 035019 ( 2013 ) ; p. m. ferreira , r. santos , h. e. haber and j. p. silva , phys .
d * 87 * , 055009 ( 2013 ) ; b. hespel , d. lopez - val and e. vryonidou , jhep * 1409 * , 124 ( 2014 ) .
n. turok and j. zadrozny , phys .
lett . * 65 * , 2331 ( 1990 ) .
a. i. bochkarev , s. v. kuzmin and m. e. shaposhnikov , phys .
d * 43 * , 369 ( 1991 ) .
a. e. nelson , d. b. kaplan and a. g. cohen , nucl .
b * 373 * , 453 ( 1992 ) .
i. f. ginzburg and i. p. ivanov , hep - ph/0312374 .
a. w. el kaffas , p. osland and o. m. ogreid , nonlin .
complex syst .
* 10 * , 347 ( 2007 ) .
i. f. ginzburg and m. krawczyk , phys . rev .
d * 72 * , 115013 ( 2005 ) .
a. barroso , p. m. ferreira , r. santos and j. p. silva , phys .
d * 86 * , 015022 ( 2012 ) . | we investigate unitarity bounds in the most general two higgs doublet model without a discrete @xmath0 symmetry nor cp conservation .
s - wave amplitudes for two - body elastic scatterings of nambu - goldstone bosons and physical higgs bosons are calculated at high energies for all possible initial and final states ( 14 neutral , 8 singly - charged and 3 doubly - charged states ) .
we obtain analytic formulae for the block - diagonalized scattering matrix by the classification of the two body scattering states using the conserved quantum numbers at high energies . imposing the condition of perturbative unitarity to the eigenvalues of the scattering matrix ,
constraints on the model parameters can be obtained .
we apply our results to constrain the mass range of the next to lightest higgs state in the model . |
multistabity is a common dynamical feature of many natural systems @xcite .
although it appears in diverse forms , a very frequently occurred variant is bistability .
there are three main manifestations of bistability : the coexistence of ( i ) two stable steady states , ( ii ) one stable limit cycle and one stable steady state , and ( iii ) two stable limit cycles .
the third form of bistability , i.e. , the coexistence of two stable limit cycles of different amplitude and frequency , generally separated by an unstable limit cycle , is called birhythmicity and oscillators showing this behavior are called birhythmic oscillators .
apart from two coexisting periodic limit cycles , birhythmicity may appear in a much more complex form , e.g. , coexistence of two chaotic attractors @xcite .
birhythmic oscillators are very common , particularly , in physics ( e.g. , energy harvesting system , see ref .
@xcite and references therein ) biology ( e.g. glycolytic oscillator and enzymatic reactions @xcite ) and chemistry @xcite .
most of the biochemical oscillations that govern the organization of cell cycle , brain dynamics or chemical oscillations are birhythmic ; examples include , birhythmicity in the p53-mdm2 network , which is the key protein module that controls proliferation of abnormal cells in mammals @xcite , intracellular ca@xmath0 oscillations @xcite , oscillatory generation of cyclic amp ( camp ) during the aggregation of the slime mold _ dictyostelium discoideum _ @xcite and circadian oscillations of the per and tim proteins in _
@xcite . in physical and engineering systems
birhythmicity plays a negative role in limiting the efficiency of a certain application .
take the practical example of an energy harvesting system that converts wind - induced vibrational energy into electrical energy .
this type of energy harvesting systems show birhythmicity @xcite , but for an efficient harvesting it is desirable that the system always resides on the large amplitude limit cycle because that produces a significant mechanical deformation , which , in turn results in larger amount of harvested electric power .
further , the presence of birhythmicity makes a system vulnerable to noise : depending upon the noise intensity the system may end up in any of the two limit cycles , which results in an unpredictable system dynamics @xcite .
therefore , monorhythmicity is of practical importance in most of the physical systems . on the other hand , in networks of neuronal oscillators
the occurrence of birhythmicity is often desirable to generate and maintain different modes of oscillations that organize various biochemical processes in response to variations in their environment @xcite .
therefore , identifying an efficient control technique is of importance that can tame birhythmicity to yield monorhythmic oscillation or can retain its character intact where ever needed .
although several mechanisms are proposed for controlling bistability consisting of oscillation and steady state @xcite ( for an elaborate recent review on the control of multistability see @xcite and references therein ) , only a few exist to control birhythmicity .
@xcite reported an effective control mechanism of birhythmic behavior in a modified van der pol system by using a variant of pyragas technique of time delay control @xcite and they showed that depending upon the time - delay one can induce monorhythmic oscillation out of birhythmicity .
but , due to the presence of time delay the system becomes infinite dimensional and thus a detailed bifurcation analysis for a wide parameter space is difficult and was not reported there .
further , the authors of @xcite established that their technique can _ suppress _ the effective birhythmic zone but can not eliminate it completely for all possible sets of nonlinear damping parameter values . in this context , another interesting control technique has been reported recently by @xcite , where the authors demonstrated that multistable systems with coexisting either periodic or chaotic attractors can be converted into a monostable one by applying an external harmonic modulation and a positive feedback to a proper _ accessible _ system parameter . in the present paper
we propose an effective and much more general control technique , that we call the conjugate self - feedback control , which is able to eliminate birhythmicity and induce monorhythmic behavior .
we consider a modified van der pol equation that has been proposed to model enzyme reactions in some biosystems @xcite and also has been studied earlier as a prototypical model that exhibits birhthmicity @xcite . with a detailed bifurcation study
we establish the effectiveness of the proposed control technique in taming birhthmicity and inducing monorhythmicity . depending upon the value of the self - feedback strength it also offers freedom to select one of the desired dynamics .
we also demonstrate our results experimentally with an electronic circuit and verify that our results are robust enough in a real - world setup where the presence of parameter fluctuation and noise are inevitable .
first we describe the model used in the following . consider a birhythmic van der pol oscillator given by @xcite @xmath1 here , @xmath2 , @xmath3 and @xmath4 are parameters that determine the nonlinear damping . in ref .
@xcite , kadji _ et al .
_ considered , @xmath5 , and by using the harmonic decomposition method they arrived at the following amplitude equation : @xmath6 equation is the generic form of the codimension - two saddle - node ( sn ) bifurcation . note that eq .
is independent of the parameter @xmath7 .
the two parameter bifurcation diagram in the @xmath8 parameter space is shown in fig .
[ ab ] ( a ) that exhibits a cusp type of codimension - two bifurcation .
the exchange of rhythmicity is through the saddle - node bifurcation of the limit cycle ( snlc ) ( shown by the solid black line in the figure ) .
[ ab](b ) shows the controlled case with @xmath9 ( @xmath10 is the control parameter to be discussed later ) , where birhythmicity is completely removed and only monorhythmicity exists .
space with @xmath11 .
bifurcation diagram for ( a ) eq .
( i.e. , without control ) , snlc : saddle - node bifurcation of limit cycle ; ( b ) with control ( for @xmath9 of eq.[vdpeqcoup ] ) : the control eliminates birhthmicity and only one limit cycle ( lc ) exists . , scaledwidth=40.0% ]
next we introduce a conjugate self - feedback term @xmath12 in eq .
@xmath13 which contains the variable of our interest , @xmath14 , and its canonical conjugate @xcite , @xmath15 ; here @xmath10 controls the strength of the self - feedback .
further , a close inspection reveals that the self - feedback mechanism effectively controls the damping of the system through the @xmath15 variable and the effective frequency through the @xmath14 variable .
however , understanding of their collective effect on the dynamics needs a detailed analysis that we will address next . to unravel the underlying dynamics of the controlled system we use the harmonic decomposition method .
let us assume the approximate solution of be given by @xmath16 with @xmath17 being the amplitude and @xmath18 the frequency of the oscillator with feedback . substituting this in yields the following expression @xmath19 but according to ref .
@xcite , we can ignore the higher harmonics regarding them as forcing term , which diminish with increasing harmonics .
thus , eq . can be reduced to @xmath20 the equation suggests the following frequency and amplitude equations , respectively , @xmath21 and @xmath22 it is interesting to note that , eq . is equivalent to eq .
when @xmath23 , i.e. , in the absence of any feedback .
also , it may be noted that the amplitude of the system depends on @xmath7 when @xmath24 , contrary to eq . .
the frequency in the harmonic limit corresponds to @xmath25 .
further , eq . imposes an upper limit on the strength of the feedback , namely @xmath26 otherwise the frequency becomes imaginary , which is non - physical .
the three roots ( actually six roots , with @xmath27 , @xmath28 . ) correspond to the amplitudes of three limit cycles ( two stable , one unstable ) . for the parameter set @xmath11 , @xmath29 , @xmath30 for different values of the coupling parameter @xmath10 .
the solid dark ( red ) curve for @xmath31 represents single limit cycle with large amplitude and the solid gray ( green ) curve for @xmath32 represents that with small amplitude . in between the curve for @xmath33
is for birhythmic oscillation .
the lower curve for @xmath34 represents stable steady state.,scaledwidth=40.0% ] now let us discuss how to determine the presence of limit cycles and their stability out of the above analytical results .
the amplitude equation eq . may be solved by graphical method .
the solutions are those for which the function @xmath49 crosses the horizontal zero line .
we consider the parameter set @xmath11 , @xmath29 and @xmath30 for which exhibits birhythmicity in the absence of self - feedback ; next , we vary the coupling strength @xmath10 to get different solutions .
the number of limit cycles is determined by the number of solutions of the amplitude equation .
the number provides the information of the steady state solution ( i.e. , no solution ) , existence of a single limit cycle ( monorhythmicity ) or three limit cycles ( birhythmicity , one of the lcs is unstable ) . from fig .
[ ava ] we find that for @xmath34 there is no zero crossing of the curve , i.e. , there is no lc and the system is in a steady state . as we decrease @xmath10 , the @xmath49 curve crosses the horizontal zero line from below and gives rise to a stable lc .
this is shown for @xmath32 with solid gray ( green ) line , here the system has only one stable lc of small amplitude .
further decrease in @xmath10 brings it to the birhythmic regime where the @xmath49 curve crosses the horizontal zero line at three different values of @xmath50 indicating three lcs ( shown for @xmath33 ) .
the stability of three lcs are determined by eq . , which suggests that the middle zero point of the curve in fig .
[ ava ] represents the unstable lc .
further increase in @xmath10 brings the system to a monorhythmic region with the large lc . the case of large single lc for @xmath9
is shown in the upper solid dark ( red ) line .
the original birhythmic van der pol oscillator given by exhibits only global snlc type of bifurcation .
however , due to the presence of the feedback term in the controlled case ( i.e. , eq.[vdpeqcoup ] ) , eq . is modified to eq . , and thus the system additionally exhibits local bifurcation , namely hopf bifurcation .
we derive the value of @xmath10 for which hopf bifurcation occurs from the eigenvalues of the jacobian of eq . around the steady state @xmath51 .
the eigenvalues are given by @xmath52 equation gives the condition of hopf bifurcation as @xmath53 where @xmath54 is the value of @xmath10 for which hopf bifurcation occurs .
space for @xmath29 , @xmath30 ,
( b ) bifurcation diagram with @xmath10 for @xmath11 ( the horizontal broken line in fig .
[ dmm](a ) ) .
sss : stable steady state .
( @xmath55 ) is the width of birhythmic zone.,scaledwidth=45.0% ] : large amplitude single lc .
( c , d ) @xmath33 : birhythmic oscillations , the blue trajectory in ( d ) shows unstable lc .
( e , f ) @xmath32 : small amplitude single lc .
( g , h ) @xmath34 : stable steady state .
the solid ( red ) line is for initial conditions @xmath56 , @xmath57 ; the dotted ( black ) line with initial condition @xmath58 , @xmath57 .
other parameters are : @xmath11 , @xmath29 , @xmath30.,scaledwidth=41.0% ]
in this section we investigate the possible bifurcation scenarios of the system using the continuation package xppaut .
we explore the nature of the bifurcation with the variation of the feedback parameter @xmath10 for different system parameters ( e.g. , @xmath7 , @xmath59 and @xmath60 ) . the bifurcation structure in the @xmath61 space is computed and shown in fig .
[ dmm](a ) .
the value of @xmath29 and @xmath30 are kept in the birhythmic zone of the uncontrolled system ( cf .
we find that the two - parameter space is divided by global bifurcations , namely saddle node bifurcation of limit cycle ( snlc ) and a local bifurcation , namely the supercritical hopf bifurcation ( hb ) . in between two snlc curves
birhythmic behavior exists [ purple ( gray ) zone ] : in this zone three lcs exist , of which two are stable ( one with smaller amplitude and the other with larger amplitude ) and an unstable lc . the transition from birhythmic to monorhythmic dynamics [ indicated by green ( light gray ) zone ] is governed by these snlc curves .
whereas the hb curve governs the transition between single stable limit cycle and stable steady state ( sss ) [ blue ( dark ) zone ] ; note that the occurrence of the hopf bifurcation agrees with our analytically predicted value of @xmath10 in .
for a clearer understanding of the bifurcation scenario we take an exemplary value @xmath11 and vary the feedback term @xmath10 [ along the broken ( yellow ) horizontal line in fig .
[ dmm](a ) ] .
the one parameter bifurcation diagram corresponding to this variation is shown in fig .
[ dmm](b ) . in the absence of the self - feedback ,
i.e. , for @xmath23 , the system is in a birhythmic zone for any @xmath2 ( in the present parametric set up ) .
if we increase @xmath10 , for @xmath62 , the system enters into the monorythmic zone via snlc bifurcation . here
we observe that the sole limit cycle in the system is the small amplitude lc .
this small lc looses its stability through an inverse hopf bifurcation and gives birth to a stable steady state . in the negative side of @xmath10 , for @xmath63
, we again have a monorhythmic region but with a large amplitude limit cycle . therefore , with a proper choice of the self - feedback strength @xmath10 one can induce monorhythmic oscillation of smaller ( @xmath62 ) or larger ( @xmath63 ) amplitude .
interestingly , a hysteresis appears around @xmath23 having a width of @xmath64 [ light gray ( purple ) of fig .
[ dmm](b ) ] . in this range of @xmath10
the system may end up showing lc of large or small amplitude depending upon initial conditions . also , the two lcs are separated by an unstable lc [ shown in dark ( blue ) line ] .
it is worth noting that the width of the hysteresis zone increases with increasing @xmath7 .
typical time series with the variation of @xmath10 are shown in fig .
[ vd ] ( @xmath11 , @xmath29 , @xmath30 ) . to detect the presence or absence of birhythmicity
, we consider a large number of initial conditions of @xmath65 .
however , here we present the results for two different initial conditions only : one around the origin ( targeting the small amplitude lc ) and the other far from the origin ( targeting the large amplitude lc ) .
the red line ( solid ) indicates the oscillation corresponding to the initial condition @xmath66 and the black line ( dotted ) indicates the oscillation for the initial condition @xmath67 .
we start from a negative @xmath10 with @xmath63 .
[ vd](a ) ( time series ) and [ vd](b ) ( phase plane plot ) show the scenario for @xmath31 .
both initial conditions result in the large amplitude lc indicating monorhythmicity .
next , we choose @xmath68 , i.e. , the birhythmic region .
figure [ vd](c ) and [ vd](d ) show this scenario for @xmath33 .
the blue trajectory in fig . [ vd](d ) indicates the unstable lc that separates the basin of attraction of two lcs , i.e. , the small lc resulted from @xmath69 and the large lc resulted from @xmath70 .
figure [ vd](e ) and [ vd](f ) show monorhtyhmic oscillation for @xmath32 ( i.e. , @xmath62 ) . here
all the initial conditions go to the smaller amplitude lc .
finally , further increase in @xmath10 results in the stable steady state [ fig .
[ vd](g ) and [ vd](h ) for @xmath34 ] . therefore ,
with the variation of @xmath10 we can effectively control the birhythmic nature of the system and can induce monorhythmic oscillation of preferred amplitude .
space for @xmath11 , @xmath30 .
the yellow broken line indicates @xmath72 where snlc and hb curves intersect .
@xmath73 is the cusp point .
sss : stable steady state , lc@xmath74sss : bistable zone with one stable steady state and one stable limit cycle .
( b ) bifurcation diagram obtained by sweeping @xmath10 along the yellow broken line of fig .
[ daa](a ) . ,
scaledwidth=43.0% ] space for @xmath11 , @xmath29 .
@xmath73 is the cusp point .
( b ) bifurcation diagram with @xmath10 for @xmath75 [ along the yellow broken line of fig .
[ dbb](a)].,scaledwidth=41.0% ] next , we investigate the effectiveness of the control over the whole nonlinear damping parameter space .
significantly , we find that one can indeed induce monorhythmicity for any set of ( @xmath59 , @xmath60 ) by choosing a proper value of @xmath10 . to systematically understand the scenario , we study the dynamics in the @xmath76 and @xmath77 space , separately . figure [ daa ] shows the two - parameter bifurcation in the @xmath76 space for @xmath30 and
figure [ dbb](a ) shows the same in the @xmath77 space for @xmath29 ( in both the cases we take @xmath11 ) . from these
two bifurcation diagrams it is seen that for @xmath78 the system has only a single lc for any choice of ( @xmath59 , @xmath60 ) ( @xmath73 is the cusp bifurcation point ) . the hb curve and the snlc curve intersect at @xmath79 ( say ) in fig.[daa ] ( a ) and at @xmath80 ( say ) in fig.[dbb](a ) .
figure [ daa](b ) shows the bifurcation scenario with the variation of @xmath10 along the horizontal broken yellow line of fig .
[ daa](a ) ( i.e. , for @xmath81 ) .
an interesting transition occurs for @xmath82 ( @xmath83 ) : if @xmath10 is increased from below , the system generates a transition from birhythmicity to another type of bistability , namely the _ coexistence of stable lc and stable steady state_. the genesis of this transition is also quite interesting .
normally , in a hysteric transition , the transition from stable steady state to stable lc occurs through a _ subcritical _
hopf bifurcation and the reverse transition occurs through a snlc @xcite , but here two snlc and one supercritical hopf bifurcation govern the hysteric transition .
this is shown in fig .
[ dbb](b ) for @xmath84 by sweeping @xmath10 along the yellow broken line of fig .
[ dbb](a ) . also note that the hopf bifurcation occurs at @xmath85 and independent of @xmath59 and @xmath60 as predicted in eq .
[ hopf ] . finally , we summarize our results in the @xmath8 parameter space . for the uncontrolled system ,
i.e. , @xmath23 , birhyhmicity occurs in a broad zone of ( @xmath8 ) values as shown in fig .
[ ab](a ) .
but , for @xmath86 the birhythmic zone is completely eliminated and the only possible dynamics is essentially monorhythmic [ fig . [ ab](a ) for @xmath9 ] .
therefore , our study reveals that a proper choice of the control parameter @xmath10 can effectively eliminate birhythmicity to establish monorhythmic oscillation and at the same time its variation may give rise to transitions between several interesting dynamical states ; by controlling @xmath10 one can achieve any of these states in a deterministic way .
experimental observation of birhythmicity is subtle due to the presence of inherent noise and parameter fluctuation in a real system and also owing to the fact that , in experiments one can record only one oscillation at a time @xcite .
the first experimental observation of birhythmicity was made by decroly and goldbeter @xcite in a chemical system , namely the parallel - coupled bromate - chlorite - iodide system . in their experiment
the time scale was of the order of few minutes .
in biological experimental setups the time scale is usually of the order of few hours , e.g. , birhythmic oscillation in the p53 system has two time scales of six and ten hours @xcite . in this context ,
the experimental observation of birhythmic oscillation in electronic circuit possesses two distinct advantages : first , the time scale is much reduced , of the order of mili second and the second one is the controllability of electronic circuits . to demonstrate birhythmicity and verify the robustness of our proposed control scheme , we realize the system given by eq . in the electronic circuit . the detailed circuit diagram is shown in fig .
[ ckt_init ] . here
m1-m4 are analog multiplier ics ( ad633jn ) and a1-a9 are opamps ( tl074 ) .
the resulting circuit equation takes the following form [ sec : expt ] the above equation becomes dimensionless for the following substitutions : @xmath89 , @xmath90 , @xmath91 , @xmath92 , @xmath93 , @xmath94 v , @xmath95 v , and @xmath96 ; with these eq . is reduced to eq . .
@xmath97 : large amplitude single lc .
( c , d ) @xmath98 @xmath97 : birhythmic oscillations .
( e , f ) @xmath99 @xmath97 : small amplitude single lc .
( g , h ) @xmath100 @xmath97 : stable steady state .
the large amplitude lc is for initial conditions @xmath101 volt , @xmath102 volt and the small amplitude lc is for initial condition @xmath103 volt , @xmath102 volt.,scaledwidth=47.0% ] we consider the following values of the used circuit components : @xmath104 k@xmath97 , @xmath105 @xmath97 , @xmath106 v and @xmath107 mv throughout the experiment .
the initial conditions are controlled through the data acquisition system ( daq ) in labview environment @xcite through a computer . to have a selected initial conditions , the capacitors ( @xmath108 ) in the integrators ( a5 and a7 )
are charged with external voltages ( @xmath109 and @xmath110 ) .
these voltages are controlled by the daq .
the voltages are connected to relays ( s1 and s2 ) to be on for a particular time period .
the on time of the relays are controlled by a microcontroller ( arduino uno @xcite ) , which is programmed to keep the relays on for a time interval of @xmath111 seconds . during this time
the capacitors @xmath108 of the integrators get charged to the desired input voltages ( @xmath109 and @xmath110 ) which are taken and controlled from the computer through the daq .
then the relays are made off and the circuit operates in its normal action .
the experimental time series and phase plane plots are shown in fig .
[ expt ] . to observe the large amplitude single lc shown in figure [ expt](a ,
b ) we add an inverter in the output terminal of a9 of fig .
[ ckt_init ] ( not shown in the figure ) and take @xmath112 @xmath97 .
[ expt](c ) and ( d ) show the scenario of birhythmicity for @xmath98 @xmath97 .
the presence of oscillations of two different amplitudes and frequencies confirms the occurrence of birhythmicity in the circuit .
the increasing @xmath113 brings the system to a monorhythmic one .
the situation for @xmath99 @xmath97 is shown in fig .
[ expt](e ) and ( f ) . with further increase in @xmath113 the oscillation
is quenched and the system rests in the stable steady state .
[ expt](g ) and ( h ) shows the case for @xmath100 @xmath97 . note the qualitative resemblance between the experimental scenarios and the numerical results of fig.[vd ] .
in summary , we have proposed a scheme to control birhythmic behavior in nonlinear oscillators .
our control scheme incorporates a self - feedback term that is governed by the variable to be controlled and its canonical conjugate .
we have considered a prototypical model that shows birhythmic oscillation and has relevance in modeling biochemical processes .
our study has revealed that a proper choice of the control parameter can effectively eliminate birhythmicity for any choice of nonlinear damping parameters and at the same time its variation may give rise to transitions between several interesting dynamical behaviors .
physical implementation of our control scheme is very much feasible , since feedback through conjugate variables is quite natural in many experimental setups @xcite .
we can realize the control scheme if we have access to at least one of the variables of interest ; from that we can always generate its time derivative via real time signal processing .
we believe that our study may have potential applications in controlling birhythmicity in several mechanical and biochemical processes as well as in other fields .
d.b . acknowledges the financial support from csir , new delhi , india , t.b .
acknowledges the financial support from serb , department of science and technology ( dst ) , india [ project grant no .
: sb / ftp / ps-005/2013 ] .
[ 1][1]#1 32ifxundefined [ 1 ] ifx#1 ifnum [ 1 ]
# 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) in @noop _ _ , , ( , , ) pp . @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , , ) @noop _ _ ( , , ) @noop * * , ( ) @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) https://books.google.co.in/books?id=2vi3rgeacaaj[__ ] ( , ) @noop _ _ , ed .
( , , ) @noop * * , ( ) | birhythmicity arises in several physical , biological and chemical systems .
although , many control schemes are proposed for various forms of multistability , only a few exist for controlling birhythmicity . in this paper
we investigate the control of birhythmic oscillation by introducing a self - feedback mechanism that incorporates the variable to be controlled and its canonical conjugate . using a detailed analytical treatment , bifurcation analysis and experimental demonstrations
we establish that the proposed technique is capable of eliminating birhythmicity and generates monorhythmic oscillation .
further , the detailed parameter space study reveals that , apart from monorhythmicity , the system shows transition between birhythmicity and other dynamical forms of bistability .
this study may have practical applications in controlling birhythmic behavior of several systems , in particular in biochemical and mechanical processes . |
type ia supernovae ( sne ia ) are observed to display remarkably homogeneous light curves in broadband photometry .
they are currently classified according to a single - parameter family of peak luminosity as a function of light curve shape @xcite , which allows them to be used as accurate cosmological distance indicators .
however , we know that this one - parameter family does not completely encompass the variety of sne ia observed .
once we have reduced any contribution from measurement error to the point where its effects are negligible the intrinsic diversity of sne ia themselves represent the limiting accuracy to which a one - parameter family can aspire .
this diversity manifests itself as differences in the sn ia spectra .
it is important to assess the diversity in particular features of sn ia spectra and find those hot spots that have the most significant effect on the broadband photometry used for cosmology .
identifying these points of variability in sn ia spectra has a multi - fold benefit .
( 1 ) it allows us to identify which spectral regions are the most stable for photometry and thus design future dedicated sn cosmology probes to utilise these stable wavelength regions .
( 2 ) it allows us to classify sn ia based on certain spectral features into sub - types that have a tighter magnitude - redshift relation than the entire sn ia population .
( 3 ) it allows us to put weighted error bars on the points in the magnitude - redshift relation depending on which part of a redshifted sn ia spectrum was sampled by the filter set in use . the quest to understand the diversity in the type ia supernova population is accelerating , with many groups investigating this in an attempt to control systematic uncertainties associated with using type ia supernovae as standard candles .
empirical methods for studying the variety in the type ia population focus on the linked aspects of photometric diversity @xcite and spectroscopic diversity .
many of the spectral studies concentrate on fitting spectral lines ( such as the si ii feature ) , measuring how they evolve with time , relating these to physical parameters such as metallically or ejecta velocity , and studying correlations between these features and the peak magnitude or light curve width @xcite .
efforts to compare these observational parameters to the results of theoretical models are ongoing ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and this important topic promises to allow the prediction of potential changes in type ia supernova demographics over time . in the meantime observational correlations , such as the width - luminosity relation , remain the strongest methods for controlling the effects of supernova diversity on cosmological parameter estimation . in this paper
we design an empirical measure of diversity that uses a new approach . instead of fitting spectral lines we subject the entire spectrum to a number of different statistical analyses .
these show promise for providing an objective method of distinguishing peculiar supernovae ( at the very least ) and for finding correlations amongst subsets of the type ia population . in section [
sect : data_analysis ] we describe the data set and our analysis technique . in section [ sect : discussion ] we provide plots of the spectral variability of sne ia as a function of wavelength and discuss the features .
in section [ sect : peculiarity ] we provide a prescription for determining the degree of ` peculiarity ' of a supernova based on its spectrum near maximum light .
the sample consists of 18 previously published spectra taken for 8 low redshift type ia sne .
table [ table : spectra ] lists the spectra and the source catalogues .
these represent the subset of published objects that are confirmed to be normal sne ia and have good quality spectra within two days of maximum light as well as established @xmath0 values , b magnitudes at maximum light ( @xmath1 ) , extinction values ( @xmath2 ) and distance moduli ( @xmath3 ) .
table [ table : objects ] lists the photometric properties of the source objects .
.catalogue of sne ia spectra used in this study.[table : spectra ] [ cols="^,>,>,>,^",options="header " , ] + @xmath4references .
br83 : branch et al . 1983 ; br03 : branch et al .
2003 ; ha02 : hamuy et al . 2002 ; kn03 : knop et al .
2003 ; le93 : leibundgut et al . 1993 ; pa96 : patat et al . 1996
; ph99 : phillips et al . 1999 ;
sah01 : saha et al . 2001 ;
sal01 : salvo et al .
2001 ; st02 : stritzinger et al . 2002
; we94 : wells et al .
1994 correction for dust extinction is made according to the standard extinction law of@xcite , using colour excess values @xmath2 from @xcite where available .
it is a familiar task to convert apparent magnitudes to absolute magnitudes in order to compare the brightness of objects at different distances . in order to compare the 18 spectra in our sample we have to perform the spectral equivalent of converting from apparent to absolute magnitude .
we do this by multiplicatively scaling the spectra so they give the correct absolute magnitude in the b or v band after synthetic photometry .
there is more than one way to determine the ` correct ' absolute magnitude .
we could choose to scale each spectrum to the broadband magnitude of its source supernova , once corrected for extinction , distance modulus and @xmath0(b ) .
however the observational uncertainty on nearby sn ia magnitudes is typically about 0.2 magnitudes , while the standard deviation in magnitude values is typically about 0.4 and 0.2 magnitudes in b and v respectively @xcite .
this makes it difficult with the current data to distinguish broadband magnitude variations due to intrinsic spectral variability from variations due to observational error .
instead we choose to scale each spectrum to a fixed broadband magnitude .
this allows us to probe the spectral diversity without contamination from observational magnitude uncertainties .
when we scale to a fixed magnitude we artificially minimise the diversity in that region .
to examine the magnitude of this effect , we scale the spectra to fixed @xmath5 in the first instance and to fixed @xmath6 in the second .
the choice of scaling has a small effect compared to the size of the spectral variations .
the mean values of the b and v absolute magnitudes are @xmath7 and @xmath8 ( @xmath9 ) , agreeing with the better performance of the @xcite relation in the v band . scaling in a single band
is achieved by allowing for a linear relationship between the synthetic photometric magnitude and the logarithm of the scaling factor ( luminosity ) and @xmath6 simultaneously , which creates a wider , more shallow dependence of variability on wavelength band , further reducing the intensity relative to the spectral variations .
this is achieved by minimising a best - fit function for fixed magnitudes @xmath10 and @xmath11 , @xmath12 differentiating with respect to @xmath13 and setting the result equal to zero yields @xmath14 the values of the coefficients are found as in the single - band case .
equation [ eq : predict ] allows us to determine _ a priori _ the scaling factor required to fix a spectrum near a chosen magnitude in both @xmath15 and @xmath16 .
] , @xmath17 the value of the slope constant @xmath18 is constant for all objects within a particular band .
the offset @xmath19 varies depending on the units of the spectrum , and is determined for each object separately by computing the relation [ eq : mk ] for a range of scaling factors @xmath20 . from this
follows the relationship for @xmath20 in terms of a fixed desired magnitude @xmath21 .
this process provides a set of extinction - corrected sn ia spectra that represents the most homogeneous spectral set we are able to generate .
we now test these spectra for diversity in the sample .
for each object we compile a single spectrum based on the average of all available spectra for that object between @xmath22 and 2 days from maximum .
we use logarithmic bin spacing to offset the doppler broadening of linewidths at higher wavelengths .
the eight average spectra are binned to constant resolution , @xmath23 , with bin size proportional to the bin centre , @xmath24 our choice of @xmath25 corresponds to approximately 20 bins near 4000 and 35 bins near 7000 .
this choice of binning parameter is narrow enough to appropriately sample all prominent spectral features , but wide enough to be insensitive to noise , allowing for effective synthesis of the spectra .
we tested a variety of bin widths from @xmath26 to @xmath27 and the results are not sensitive to the width chosen . using the averaged spectra from the eight supernovae in our sample we calculate the overall mean spectrum and determine the root - mean - square ( rms ) deviation from the mean in each bin to quantify the diversity in that region .
the result is plotted in figures [ fig : scatter : fixedb ] and [ fig : scatter : fixedv ] , where the lower panel shows the overall average spectrum and the solid line in the middle panel shows the rms dispersion about the mean .
the shaded regions in this figure represent the uncertainty in the rms dispersion , which is estimated by bootstrap resampling @xcite .
we wish to isolate the intrinsic diversity of sne ia from observational scatter . as the spectra used in the sample range from epoch -2 to 2 days we expect some diversity within the spectra for a single object .
moreover , differences between individual observations introduce scatter that is independent of intrinsic diversity in the sn ia population .
we seek to quantify this variance and remove it from the total , leaving only the intrinsic rms .
we have already made a single spectrum for each object by taking the average of all its available spectra . for each object
, we account for the spurious dispersion by calculating the rms variance of the available spectra for each object about the mean of that object .
this quantity the intra - object variation when averaged over all the objects , gives an estimate of the dispersion that is not due to intrinsic differences between the supernovae .
this quantity is also used as an estimate of the non - intrinsic dispersion for an object when only one spectrum is available .
this dispersion is the dashed line in the middle panel of figures [ fig : scatter : fixedb ] and [ fig : scatter : fixedv ] .
it is small compared to the dispersion between objects . since the contribution from observational error
is expected to only increase the scatter in the results we subtract ( in quadrature ) the observational scatter component from the overall rms deviation .
the remaining rms , ideally all due to intrinsic diversity , is shown in the upper panel of figures [ fig : scatter : fixedb ] and [ fig : scatter : fixedv ] .
this method allows us to separate spectral diversity from scatter in absolute magnitudes
. it will be sensitive to diversity in spectral features as well as colour variaions large scale spectral shape differenes .
the band to which we scale the spectrum will have the least diversity because we are effectively anchoring the dispersion to be lowest at this point .
so when we are scaling to a fixed @xmath1 the scatter is smallest between 4000 and 5000 , whereas when scaling to @xmath28 the scatter is smallest between 5000 and 6000 .
the change in the overall shape of the dispersion spectrum is slight , and the significant spectral features remain unaltered . in figure
[ fig : fixedzoom ] we compare the final results for scaling to the v band ( upper panel ) and b band ( lower panel ) . in figure
[ fig : sniamax ] we show a typical sn ia spectrum at maximum light @xcite .
the spectrum peaks at 4000 .
our analysis shows that blueward of this peak the uv edge of the spectrum shows more variety than the rest of the spectrum .
much of this is likely to be intrinsic sn ia diversity , but it is also the region most susceptible to extinction and to observational flux errors . rejecting the blue end of sn ia spectra will reduce the scatter in sn ia magnitudes that occurs both due to intrinsic sn variability and imperfect extinction correction .
the rayleigh - jeans tail to the right of this peak , is a very stable region of the spectrum until the si ii feature around 6150 produces a peak in variability . beyond the si
ii feature the diversity also appears to increase .
however , this may be another indication of imperfect extinction correction .
we expect the dominant source of observational error that can not be accounted for by measuring intra - object dispersion to be due to extinction correction . by sampling the averaged object spectra with replacement many times we estimate the error in the extinction correction without making any assumptions about the error in colour excess values .
because we anchor the spectra to the b or v magnitudes of the supernovae ( even when we are anchoring to the b or v magnitude of each supernova individually ) , any faulty extinction correction would show up as a bowl - shaped curve on our rms vs wavelength plots ( affecting high and low wavelengths equally - even though extinction preferentially affects short wavelengths ) .
the curves in figure [ fig : fixedzoom ] may be the most illustrative for this point , since we have scaled the spectra to a fixed magnitude in b and v ( upper and lower ) , effectively minimising the scatter in these regions .
an error in extinction correction shows up as a gradual increase in the scatter away from these regions . in this analysis
we are particularly concerned with ( and sensitive to ) small scale variation such as the variabiliy of a particular spectral feature .
prominent spectral features correspond to regions of high scatter , most noticably around the si ii line at 6150 .
ideally we would avoid these regions when calculating the magnitudes of sne ia for use as standard candles .
however , a large variation in a spectral line does not necessarily translate into a large variation in broad - band magnitude if the regions around the line are bright and stable , in which case the variability of these lines will find the greatest utility as a tracer of different subsets of the sn ia population .
in any real experiment we have a limited filter set , so can not choose to observe the ideal region of every sn spectrum for the continuum of redshifts we need to cover .
however , we can weight the error bars of our observations to favour those objects whose redshifts fortuitously allow one of our filters to cover the most stable region , and we can design filters narrow enough , and with sufficient overlap , to maximise the chance that a filter will cover only the most stable region @xcite
. our analysis does not take into account any spectral differences between supernovae with different @xmath0 values .
when enough data is available it will be straightforward to apply this analysis to identify spectral features associated with wide or narrow light curves .
nugent / nugent_templates.html . here
we overplot our average spectrum on the nugent template for a sn ia at maximum light .
the agreement is good .
our analysis shows that the uv edge of the spectrum , below the peak around 4100 , shows an increased diversity among sne ia .
( whether this continues below the limit of our useful data , at 3700 , is an open question . )
this also approximately corresponds to where our average spectrum starts to deviate significantly from the nugent template - corresponding to the greater uncertainty about the typical spectrum of this region .
we have shown that the rayleigh - jeans tail to the right of the peak is a very stable region of the spectrum , until we reach the si ii feature at 6150 .
this is therefore the best region of the spectrum to use as a standard candle for cosmological parameter estimation . ]
when using sne ia for cosmology we need to select the observed supernovae with the most homogeneous characteristics for use as standardisable candles .
most well observed supernovae exhibit some peculiar features , and their full utility in cosmology requires the characterisation of these differences .
a significant step toward this is the creation of an objective measure of peculiarity that can act as an objective criterion for including an object in a sample population , and also identify subsets of sne ia with very similar properties .
here we outline two different methods for quantifying peculiarity and demonstrate them on the small data set currently available to us .
these methods will benefit greatly from a much larger low-@xmath29 spectroscopic sample , such as is now being collected by a number of groups , including the nearby supernova factory , the canada france hawaii telescope legacy survey ( cfhtls ) , the european supernova collaboration and the carnegie supernova project @xcite . in addition , applying these methods to the high-@xmath29 samples being collected by the essence collaboration , the cfhtls and the sn cosmology project will allow for tests of evolution in the demographics of the sn ia population .
the first method uses a @xmath29-test to determine the degree to which a sn ia deviates from the norm , and is particularly good at finding spectra with variations in regions which are usually stable , or spectra with strange continuum profiles .
this means it is also good at picking out dubious extinction corrections .
the second method uses principle component analysis and is particularly good at finding subsets of the population with similar characteristics .
for example , it can pick out sne ia with high or low ejecta velocities by emphasising correlated differences between spectral features .
we use our calculation of intrinsic diversity to give a measure of the range within which supernovae can be considered normal at each wavelength .
if an object deviates significantly from the average in a region of the spectrum which is very stable , such as the range from 4000 to 6000 , we might consider it ` peculiar ' .
this can be quantified by assigning a @xmath29-score in each wavelength bin and then averaging the @xmath29-scores over some region , e.g. the entire spectrum or a band such as b or v. the @xmath29-score at wavelength @xmath30 is , @xmath31 where @xmath32 , @xmath33 and @xmath34 are the flux of the object , the flux of the average spectrum and the dispersion of the average spectrum , all at the wavelength @xmath30 .
we define the ` peculiarity parameter ' , @xmath35 , to be the absolute average of these terms over the @xmath36 bins in the desired wavelength range , @xmath37 subsets of this range , weighted by filter value , can be used to define the broadband peculiarity parameters , e.g. for @xmath15 , @xmath38 where @xmath39 is the throughput of the @xmath15 band filter at wavelength @xmath30 . figure [ fig : pvalues_b ] shows a histogram of @xmath35 for the objects in our sample calculated according to this method using the wavelength range 3700 to 7100 , based on the object averages for which the spectra were anchored in the b band . for this test ,
the average and rms spectra are recalculated including observations from sn 1991bg and sn 1991 t , previously classified as ` peculiar ' .
the average spectrum changes negligibly when these objects are added .
the peculiarity of 1991bg is made obvious by the test , from which we conclude that the spectrum of this object departs markedly from the mean in a non - uniform ( across @xmath30 ) manner .
however sn 1991 t , which is known to be bolometrically overluminous , shows little spectral peculiarity when scaled in magnitude to match other events . .
spectra have been scaled to the average absolute magnitude in the b band .
the average spectrum from which deviation is measured is composed of average spectra from the objects listed excluding 1991bg and 1991 t . ] , with the average spectrum additionally excluding 1999ee .
spectra have been scaled to the average absolute magnitude in the b band . ] to test the fidelity of this parameter , the histogram is compiled again using instead the spectra anchored in the v band .
figure [ fig : pvalues_v ] shows the recompiled histogram .
some slight rearranging occurs notably the objects seem less bunched however the prominence of 1991bg , and the relative lack of prominence of 1991 t , are unchanged .
we conclude that this parameter can distinguish between events that exhibit peculiar broadband and spectral features , such as 1991bg - like events , but does not distinguish objects like 1991 t that are peculiar at wavelengths that also show significant diversity in normal supernovae . principle component analysis ( pca ) is a technique for identifying common characteristics in data with many variables ( e.g. * ? ? ?
pca has been successfully applied to astronomical spectra in the past to identify different types of active galactic nuclei @xcite . here
we demonstrate the concept on type ia supernova spectra .
this technique may be used in the future , with a larger sample set , as another method to identify subclasses of type ia supernovae with similar properties .
this is a major aim of several ongoing low - redshift supernova surveys and is important for the control of systematic errors in planned ( and ongoing ) high - redshift surveys that aim to measure the expansion history of our universe .
we use the pca technique developed by @xcite including their analysis code with some minor variations .
we perform pca on the corrected sample of normal supernovae as outlined in table [ table : spectra ] , as well as extinction - corrected spectra for the peculiar sne 1991 t and 1991bg .
the result is shown in fig .
[ fig : pcaallsame ] , where we have plotted cofficients of the first and second principle components , calculated for each supernova .
this test successfully identifies these two supernovae as peculiar .
moreover , this method is able to distinguish 1991bg - like sne from 1991t - like sne because they each appear peculiar in a different principle component .
this method therefore shows great promise for selecting sub - classes of sne ia .
future work will perform this analysis on multi - epoch spectra , in which sub - class trends should be even more prominent .
we have used published spectral data to calculate the diversity of type ia supernova spectra close to maximum light as a function of wavelength between 3700 and 7100 . while results in the regions toward the uv and nir are sensitive to systematics of our calibration methodology , the diversity is particularly stable between 4100 and 6000 in the rest frame of the supernova .
this is therefore the ideal region for calculating sn ia magnitudes for use as standard candles .
there is an increase in supernova diversity below 4100 .
as we move blueward , extinction correction also becomes more uncertain and therefore the region below rest frame 4100 should be avoided when using sn ia as standardizable candles .
observations of this region remain useful for constraining theoretical models of sn ia explosions , for example @xcite showed the u band is a powerful tool for discriminating between different processes occurring in sne ia .
redward of the stable region the si ii line , typically observed at 6150 , shows a large diversity .
this feature has previously been used as a tracer of light - curve width @xcite and to identify peculiar sne ia @xcite .
our analysis confirms that it is a variable feature of ordinary sne ia .
future work will extend this study to spectra over a wider epoch range .
we also demonstrated two methods for determining the peculiarity of type ia supernovae , firstly by comparing their spectra to the average sn ia spectrum while taking into account the typical variability of different spectral regions , and secondly using principle component analysis .
both methods show promise for use as objective tests of peculiarity .
pca in particular may be a good way to objectively define sub - types of the sn ia class .
the methods demonstrated here will benefit greatly from more spectroscopic data , not only more objects , but also extended wavelength coverage into the uv and ir .
although hundreds of supernovae have now been observed very well photometrically , the sample that have spectra published near maximum remains small .
new data soon to be available from a variety of searches will enable a more robust analysis based on the methods demonstrated here .
such an analysis would aim to understand the nature , rather than the extent , of the diversity in the population .
potential sources include variations in expansion velocities , chemical composition and _ ex - situ _ factors such as evolution and gravitational lensing .
relationships between separate supernova events could then be drawn by a study of correlations in spectral residuals .
this subject will be tackled in future work .
many thanks are due to paul francis for sharing with us his principle component analysis code .
we also thank the many observers who are responsible for obtaining this data .
jbj was supported by an australian national university summer research scholarship .
tmd acknowledges the support of lawrence berkeley laboratory .
we thank the anonymous reviewer for their helpful comments .
99 benetti , s. et al .
2004 , mnras , 348 , 261 benetti , s. et al .
2005 , apj , 623 , 1011 bongard , s. et al . 2006 , astro - ph/0512229 blondin , s. et al . 2006 ,
aj , in press , astro - ph/0510089 branch , d. et al . 1983 , apj , 270 , 123 ( br83 ) branch , d. , van der bergh , s. 1993 , aj , 105 , 2231 branch , d. , fisher , a. , & nugent , p. 1993 , aj , 106 , 2383 branch , d. et al .
2003 , aj , 126 , 1489 ( br03 ) branch , d. , baron , e. , hall , n. , melakayil , m. and parrent , j. 2005 , pasp , 117 , 545 branch , d. et al .
2006 , astro - ph/0601048 davis , t. m. , schmidt , b. p. , & kim , a. 2006 , pasp , _ in press _ efron , b. , 1982 , _ the jackknife , the bootstrap and other resampling plans _
, cbms - nsf regional conference series in applied mathematics , philadelphia : society for industrial and applied mathematics ( siam ) filippenko , a. v. 1997 , araa , 35 , 309 francis , p. j. , hewett , p.c . , foltz , c.b . &
chaffee , f.h .
1992 , apj , 398 , 476 francis , p. j. & wills , b. j. 1999 , quasars and cosmology , asp conference series 162 , eds ferland , g. & baldwin , j. , p. 363 gallagher , j. s. et al .
2005 , apj , 634 , 210 hamuy , m. , phillips , m. m. , schommer , r. a. , suntzeff , n. b. , maza , j. , & aviles , r. 1996 , aj , 112 , 2391 hamuy , m. et al .
2002 , aj , 124 , 417 hamuy , m. et al .
2005 , pasp , in press jha , s. et al .
2005 , aj , 131 , 527 kirshner , r. p. , et al .
1993 , apj , 415 , 589 ( ki93 ) knop , r. a. et al .
2003 , apj , 598 , 102 ( kn03 ) leibundgut , b. et al .
1993 , aj , 105 , 301 ( le93 ) nugent , p. , phillips , m. , baron , e. , branch , d. and hauschildt , p. 1995
, apjl , 455 , l147 nugent , p. , kim , a. , & perlmutter , s. 2002 , pasp , 114 , 803 patat , f. , benetti , s. , cappellaro , e. , danziger , i. j. , della valle , m. , mazzali , p. a. , & turatto , m. 1996 , mnras , 278 , 111 ( pa96 ) phillips , m. m. 1993 , apjl , 413 , l105 phillips , m. m. , lira , p. , suntzeff , n. b. , schommer , r. a. , hamuy , m. , & maza , j. 1999 , aj , 118 , 1766 ( ph99 ) saha , a. , sandage , a. , tammann , g. a. , dolphin , a. e. , christensen , j. , panagia , n. , & macchetto , f. d. 2001 , apj , 562 , 314 ( sah01 ) salvo , m. e. , cappellaro , e. , mazzali , p. a. , benetti , s. , danziger , i. j. , patat , f. , & turatto , m. 2001 , mnras , 321 , 254 ( sal01 ) savage , b. d. , & mathis , j. s. 1979 , ara&a , 17 , 73 stritzinger , m. et al . 2002 ,
aj , 124 , 2100 ( st02 ) travaglio , c. , hillebrandt , w. and reinecke , m. 2006 , aap , in press , astro - ph/0507510 wells , l. a. et al .
1994 , aj , 108 , 2233 ( we94 ) | we use published spectroscopic and photometric data for 8 type ia supernovae to construct a dispersion spectrum for this class of object , showing their diversity over the wavelength range 3700 to 7100 .
we find that the b and v bands are the spectral regions with the least dispersion , while the u band below 4100 is more diverse .
some spectral features such as the si line at 6150 are also highly diverse .
we then construct two objective measures of ` peculiarity ' by ( i ) using the deviation of individual objects from the average sn ia spectrum compared to the typical dispersion and ( ii ) applying principle component analysis .
we demonstrate these methods on several sne ia that have previously been classified as peculiar .
[ firstpage ] supernovae : general |
the 2013 nobel prize in physics was awarded to francois englert and peter higgs for their work in 1964 that led in 2012 to the discovery at the cern large hadron collider ( lhc ) of the higgs boson after its being sought for almost 50 years .
it is great tragedy that robert brout , the joint author with francois englert of one of the papers that led to the 2013 nobel prize , died in 2011 , just one year before the discovery of the higgs boson and two years before the awarding of the nobel prize for it .
( for me personally this is keenly felt since my first post - doc was with robert and francois in brussels 1970 - 1972 . )
the paper coauthored by englert and brout appeared on august 31 , 1964 in physical review letters @xcite after being submitted on june 26 , 1964 and was two and one half pages long .
higgs wrote two papers on the topic .
his first paper appeared on september 15 , 1964 in physics letters @xcite after being submitted on july 27 , 1964 and was one and one half pages long , and the second paper appeared on october 19 , 1964 in physical review letters @xcite after being submitted on august 31 , 1964 and was also one and one half pages long .
thus a grand total of just five and one half pages .
the significance of the higgs mechanism introduced in these papers and the higgs boson identified by higgs is that they are tied in with the theory of the origin of mass , and of the way that mass can arise through collective effects ( known as broken symmetry ) that only occur in systems with a large number of degrees of freedom .
such collective effects are properties that a system of many objects collectively possess that each one individually does not the whole being greater than the sum of its parts .
a typical example is temperature .
a single molecule of @xmath0 does not have a temperature , and one can not tell if it was taken from ice , water or steam .
these different phases are collective properties of large numbers of @xmath0 molecules acting in unison . moreover , as one changes the temperature all the @xmath0 molecules can act collectively to change the phase ( freezing water into ice for instance ) , with it being the existence of such phase changes that is central to broken symmetry .
i counted at least 19 times that nobel prizes in physics have in one way or another been given for aspects of the problem : dirac ( 1933 ) ; anderson ( 1936 ) ; lamb ( 1955 ) ; landau ( 1962 ) ; tomonaga , schwinger , feynman ( 1965 ) ; gell - mann ( 1969 ) ; bardeen , cooper , schrieffer ( 1972 ) ; richter , ting ( 1976 ) ; glashow , salam , weinberg ( 1979 ) ; wilson ( 1982 ) ; rubbia , van der meer ( 1984 ) ; friedman , kendall , taylor ( 1990 ) ; lee , osheroff , richardson ( 1996 ) ; t hooft , veltman ( 1999 ) ; abrikosov , ginzburg , leggett ( 2003 ) ; gross , politzer , wilczek ( 2004 ) ; nambu , kobayashi , maskawa ( 2008 ) ; englert , higgs ( 2013 ) ; kajita , mcdonald ( 2015 ) . and
this leaves out anderson who made major contributions to collective aspects of mass generation and yang who ( with mills ) developed non - abelian yang - mills gauge theories but got nobel prizes ( anderson 1977 , yang 1957 ) for something else . in this paper
i provide a pedagogical review the work of the work that led up to the discovery of the higgs boson and discuss its implications @xcite .
as introduced by newton mass was mechanical .
the first ideas on dynamical mass were due to poincare ( poincare stresses needed to stabilize an electron all of whose mass came from its own electromagnetic self - energy according to @xmath1 ) .
however this was all classical . with quantum field theory ,
the mass of a particle is able to change through self interactions ( radiative corrections to the self - energy
lamb shift ) to give @xmath2 , or through a change in vacuum ( bardeen , cooper , schrieffer bcs theory ) according to @xmath3 where @xmath4 is the self - consistent gap parameter .
then through nambu ( 1960 ) and goldstone ( 1961 ) the possibility arose that not just some but in fact all of the mass could come from self interaction , and especially so for gauge bosons , viz .
anderson ( 1958 , 1963 ) , englert and brout ( 1964 ) , higgs ( 1964 ) , guralnik , hagen , and kibble ( 1964 ) .
this culminated in the weinberg ( 1967 ) , salam ( 1968 ) , and glashow ( 1961 , 1970 ) renormalizable @xmath5 local gauge theory of electroweak interactions , and the confirming discoveries first of weak neutral currents ( 1973 ) , then charmed particles ( 1974 ) , then the intermediate vector bosons of the weak interactions ( 1983 ) , and finally the higgs boson ( 2012 ) .
all of this is possible because of dirac s hilbert space formulation of quantum mechanics in which one sets @xmath6 , with the physics being in the properties of the states @xmath7 .
we thus live in hilbert space and not in coordinate space , and not only that , there is altogether more in hilbert space than one could imagine , such as half - integer spin and collective macroscopic quantum systems such as superconductors and superfluids .
in this hilbert space we find an infinite dirac sea of negative energy particles .
this large number of degrees of freedom can act collectively to provide the dynamics needed to produce mass generation and the higgs boson .
the discovery of the higgs boson was announced by cern on july 4 , 2012 , accompanied by simultaneous announcements by the experimental groups atlas and cms at the large hadron collider at cern , as then followed by parallel publications that were simultaneously submitted to physics letters b on july 31 , 2012 and published on september 17 , 2012 , one publication by the atlas collaboration @xcite , and the other by the cms collaboration @xcite .
the atlas paper had 2924 authors , and the cms paper had 2883 .
the higgs boson signature used by both collaborations was to look for the decay into lepton pairs of any higgs boson that might be produced in high energy proton proton collisions at the large hadron collider
. from amongst a set of @xmath8 proton - proton collisions produced at the large hadron collider , of the order of 240,000 collisions produce a higgs boson . of them just 350 decay into pairs of gamma rays and just 8 decay into a pair of leptons .
the search for the higgs boson is thus a search for some very rare events .
thus to see higgs bosons one needs an energy high enough to produce them and the sensitivity to see such rare decays when they are produced . in searches over the years
it was not known in what energy regime to look for higgs particles , with the large hadron collider proving to be the collider whose energy was high enough that one could finally explore in detail the 125 gev energy domain where the higgs boson was ultimately found to exist .
in order to characterize macroscopic ordered phases in a general way landau introduced the concept of a macroscopic order parameter @xmath9 . for
a ferromagnet for instance @xmath9 would represent the spontaneous magnetization @xmath10 and would thus be a matrix element of a field operator in an ordered quantum state that described the ordered magnetic phase .
building on this approach ginzburg and landau @xcite wrote down a lagrangian for such a @xmath9 for a superconductor , with kinetic energy @xmath11 and potential @xmath12 , where @xmath13 is the critical temperature . for temperatures above the critical temperature
the potential would have the shape of a single well , viz .
like the letter u , with the coefficient of the @xmath14 term being positive . for temperatures below the critical temperature the potential would have the shape of a double well , viz
. like the letter w , with the coefficient of the @xmath14 term being negative . above the critical temperature the order parameter would be zero at the minimum of the potential ( normal phase with state vector @xmath15 in which @xmath16 ) .
below the critical temperature the order parameter would be nonzero ( superconducting state @xmath17 in which @xmath18 is nonzero ) . in 1957
bardeen , cooper , and schrieffer @xcite developed a macroscopic theory of superconductivity ( bcs ) based on cooper pairing of electrons in the presence of a filled fermi sea of electrons , and explicitly constructed the state @xmath17 .
in this state the matrix element @xmath19 was equal to a spacetime independent function @xmath4 , the gap parameter , which led to a mass shift to electrons propagating in the superconductor of the form @xmath3 .
the gap parameter @xmath4 would be temperature dependent ( @xmath20 ) , and would only be nonzero below the critical temperature . in 1959
gorkov @xcite was able to derive the ginzburg - landau lagrangian starting from the bcs theory and identify the order parameter as @xmath21 where @xmath22 is a coherent state in the hilbert space based on @xmath17 . in the superconducting case
then @xmath9 is not itself a quantum - field - theoretic operator ( viz .
a q - number operator that would have a canonical conjugate with which it would not commute ) but is instead a c - number matrix element of a q - number field operator in a macroscopic coherent quantum state . in 1958
anderson @xcite used the bcs theory to explain the meissner effect , an effect in which electromagnetism becomes short range inside a superconductor , with photons propagating in it becoming massive .
the effect was one of spontaneous breakdown of local gauge invariance , and was explored in detail by anderson @xcite and nambu @xcite . in parallel with these studies
nambu @xcite , goldstone @xcite , and nambu and jona - lasinio @xcite explored the spontaneous breakdown of some continuous global symmetries and showed that collective massless excitations ( goldstone bosons ) were generated , and that the analog gap parameter would provide for dynamically induced fermion masses . in 1962 goldstone , salam , and weinberg @xcite showed that there would always be massless goldstone bosons in any lorentz invariant theory in which a continuous global symmetry was spontaneously broken .
while one could avoid this outcome if the symmetry was also broken in the lagrangian , as was thought to be the case for the pion , a non - massless but near goldstone particle ( i.e. one with broken symmetry suppressed couplings to matter at low energies ) , in general the possible presence of massless goldstone bosons was a quite problematic outcome because it would imply the existence of non - observed long range forces . in 1962
schwinger @xcite raised the question of whether gauge invariance actually required that photons be massless , and noted for the inverse photon propagator @xmath23 that if the vacuum polarization @xmath24 had a massless pole of the form @xmath25 , then @xmath26 would behave as the massive particle @xmath27 .
a massless goldstone boson could thus produce a massive vector boson .
with anderson having shown that a photon would become become massive in a superconductor , there was a spirited discussion in the literature ( anderson @xcite , klein and lee @xcite , gilbert @xcite ) as to whether an effect such as this might hold in a relativistic theory as well or whether it might just have been an artifact of the fact that the bcs theory was non - relativistic . with the work of englert and brout @xcite and higgs @xcite , and then guralnik , hagen , and kibble @xcite , the issue was finally resolved , with it being established that in the relativistic case the goldstone theorem did not in fact
hold if there was a spontaneous breakdown of a continuous local theory , with the would - be goldstone boson no longer being an observable massless particle but instead combining with the initially massless vector boson to produce a massive vector boson .
technically , this mechanism should be known as the anderson , englert , brout , higgs , guralnik , hagen , kibble mechanism , and while it has undergone many name variations over time , it is now commonly called the higgs mechanism .
what set higgs work apart from the others was that in his 1964 physical review letters paper higgs noted that as well as a massive gauge boson there should also be an observable massive scalar boson , this being the higgs boson . at the time of its development in 1964
there was not much interest in the higgs mechanism , with all of the englert and brout , higgs , and guralnik , hagen and kibble papers getting hardly any citations during the 1960s at all .
the primary reason for this was that at the time there was little interest in non - abelian yang - mills gauge theories in general , broken or unbroken , and not only was there no experimental indication at all that one should consider them , it was not clear if yang - mills theories were even quantum - mechanically viable .
all this changed in the early 1970s when t hooft and veltman showed that these theories were renormalizable , and large amounts of data started to point in the direction of the relevance of non - abelian gauge theories to physics , leading to the @xmath28 picture of strong , electromagnetic and weak interactions , which culminated in the discoveries of the @xmath29 , @xmath30 and @xmath31 intermediate vector bosons ( 1983 ) with masses that were generated by the higgs mechanism , and then finally the higgs boson itself ( 2012 ) .
what gave the higgs boson the prominence that it ultimately came to have was the realization that in the electroweak @xmath5 theory the higgs boson not only gives masses to the gauge bosons while maintaining renormalizability , but through its yukawa couplings to the quarks and leptons of the theory it gives masses to the fermions as well .
the higgs boson is thus responsible not just for the masses of the gauge bosons then but for the masses of all the other fundamental particles as well , causing it to be dubbed the god particle " @xcite .
plotted as a function of @xmath9,width=480,height=288 ] even though broken symmetry is an intrinsically quantum - mechanical phenomenon , one can illustrate many of the aspects needed to understand the higgs boson by considering properties not of a q - number quantum field such as a quantum scalar field @xmath32 but of its c - number vacuum matrix element @xmath33 .
below we shall explore the quantum mechanics of broken symmetry , but for pedagogical purposes it is convenient to first proceed as though one could treat the problem classically by studying properties of the c - number @xmath34 without needing to delve into its origin or significance . thus consider a real classical scalar field ( just one degree of freedom ) with a potential energy in the form of the double - well potential shown in fig .
( 1 ) , viz . a potential
shaped like a letter w with two wells : @xmath35 this potential has a discrete symmetry under @xmath36 , with its first two derivatives being given by @xmath37 the potential has a local maximum at @xmath38 where @xmath39 is zero and @xmath40 is negative , and two - fold * degenerate * global minima at @xmath41 and @xmath42 where @xmath43 is equal to @xmath44 and @xmath45 is positive . since @xmath38 is a local maximum , if we consider small oscillations around @xmath38 of the form @xmath46 we generate a negative quadratic term @xmath47 and thus negative mass squared , viz .
@xmath48 , a so - called tachyon .
the tachyon signals an instability of the configuration with @xmath38 ( i.e. we roll away from the top of the hill )
. however if we fluctuate around either global minimum ( i.e. we oscillate in the vertical around the bottom of either hill ) by setting @xmath49 we get @xmath50 the field @xmath51 now has the positive @xmath52 ( @xmath53 ) .
thus ( goldstone 1961 ) the would - be tachyonic particle becomes a massive particle , and with one scalar field we obtain one particle .
this particle is the * higgs boson * in embryo .
the @xmath44 potential term contributes to the * cosmological constant * , and would be unacceptably large ( @xmath54 times too large ) if the boson has the 125 gev mass that the higgs boson has now been found to have .
all this arises because the minimum is two - fold degenerate , and picking either one breaks the symmetry spontaneously , since while one could just as equally be in either one or the other , one could not be in both .
this is just like people at a dinner .
each one can take the cup to their left or their right , but once one person has done so , the rest have no choice .
however a person at the opposite end of the table may not know what choice was made at the other end of the table and may make the opposite choice of cup , and thus persons in the middle could finish up with no cup . to ensure that this does not happen
we need long range correlations hence massless goldstone bosons .
consider a complex scalar field ( two degrees of freedom ) @xmath55 , @xmath56 , with a potential energy in the shape of a rotated letter w or a broad - brimmed , high - crowned mexican hat : @xmath57 the potential has a continuous global symmetry of the form @xmath58 with constant @xmath59 , with derivatives @xmath60 this potential has a local maximum at @xmath61 where @xmath39 is zero , and infinitely * degenerate * global minima at @xmath62 ( the entire 360 degree valley or trough between the brim and the crown of the mexican hat ) .
again we would have a tachyon if we expand around the local maximum , only this time we would get two . so fluctuate around any one of the global minima by setting @xmath63 , @xmath64 , to get @xmath65 the ( embryonic ) higgs boson field is now @xmath66 with @xmath52 .
however , the field @xmath67 no has * no mass at all * ( it corresponds to horizontal oscillations along the valley floor ) , and is known as a * goldstone boson * ( goldstone 1961 ) .
thus from a complex scalar field we obtain two particles .
since the goldstone boson is massless , it travels at the speed of light .
it is thus intrinsically relativistic , and being massless can provide for long range correlations .
moreover , if such particles exist then they could generate fermion masses entirely dynamically ( nambu 1960 , nambu and jona - lasinio 1961 ) , with the pion actually serving this purpose .
the @xmath44 potential term remains and the cosmological constant problem is just as severe as before .
now if we have massless particles , we would get long range forces ( just like photons ) , and yet nuclear and weak forces are short range .
so what can we do about such goldstone bosons .
two possibilities - they could get some mass because the symmetry is not exact ( pion mass ) , or we could get rid of them altogether ( the higgs mechanism ) .
consider the model studied by higgs ( 1964 ) consisting of a complex scalar field ( two degrees of freedom ) coupled to a massless vector gauge boson ( another two degrees of freedom ) for a total of four degrees of freedom in all , viz .
@xmath68 , @xmath69 , @xmath70 .
the system has kinetic energy @xmath71 and potential energy @xmath72 : @xmath73 and because of the gauge boson the system is now invariant under continuous * local * gauge transformations of the form @xmath74 , @xmath75 with spacetime dependent @xmath76 .
with derivatives @xmath77 the potential has a local maximum at @xmath78 where @xmath79 is zero and * degenerate * global minima at @xmath80 ( infinitely degenerate since independent of @xmath81 ) .
again we would have two tachyons if we expand around the local maximum .
so fluctuate around the global minimum by setting @xmath82 , @xmath83 .
on defining @xmath84 we obtain @xmath85 there is again a higgs boson field @xmath66 with @xmath52 .
however , the field @xmath67 has disappeared completely .
instead the vector boson now has a * nonzero mass * given by @xmath86 . since a massive gauge boson has three degrees of freedom ( two transverse and one longitudinal ) while a massless gauge boson such as the photon only has two transverse degrees of freedom , the would - be massless goldstone boson is absorbed into the now massive gauge boson to provide its needed longitudinal degree of freedom . hence a massless goldstone boson and a massless gauge boson
are replaced by one massive gauge boson , with two long - range interactions being replaced by one short range interaction .
this is known as the higgs mechanism ( 1964 ) though it was initially found by anderson ( 1958 ) in his study of the meissner effect in superconductivity .
the remaining fourth of the original four degrees of freedom becomes the massive higgs boson , and its presence is an indicator that the higgs mechanism has taken place .
however , the cosmological constant problem remains as severe as before .
to discuss broken symmetry in quantum field theory it is convenient to introduce local sources . in the gell - mann - low adiabatic switching procedure one introduces a quantum - mechanical lagrangian density @xmath87 of interest , switches on a local source @xmath88 for some field @xmath32 at time @xmath89 and switches it off at @xmath90 .
while the source is active the lagrangian of the theory is given by @xmath91 . before the source
is switched on the system is in the ground state @xmath92 of the hamiltonian density @xmath93 associate with @xmath87 , and after the source is switched off the system is in the state @xmath94 . while @xmath92 and @xmath92 are both eigenstates of @xmath93 , they differ by a phase , a phase that is fixed by @xmath88 according to @xmath95|\omega_j\rangle = e^{iw(j ) } , \label{m10}\end{aligned}\ ] ] with this expression serving to define the functional @xmath96 .
as introduced , @xmath96 serves as the generator of the connected @xmath97 theory green s functions @xmath98|\omega_0\rangle$ ] according to @xmath99 on fourier transforming the green s functions , we can expand @xmath96 about the point where all momenta vanish , to obtain @xmath100 , \label{m12}\end{aligned}\ ] ] with the physical significance of @xmath101 ( which is spacetime independent if @xmath102 is ) being that it is the energy density difference @xmath103 given @xmath96 , via functional variation we can construct the so - called classical ( c - number ) field @xmath104 @xmath105 and the effective action functional @xmath106 where the @xmath107 are the one - particle - irreducible , @xmath108 , green s functions of @xmath32 .
functional variation of @xmath109 then yields @xmath110 to relate @xmath111 back to the source @xmath102 . on expanding around the point where all external momenta vanish ,
we can write @xmath109 as @xmath112 . \label{m17}\end{aligned}\ ] ] the quantity @xmath113 is known as the effective potential ( which is spacetime independent if @xmath114 is ) , while the @xmath115 term serves as the kinetic energy of @xmath114 .
the @xmath116 green s functions can contain two kinds of contributions , tree approximation graphs that involve vertex interactions but no loops , and radiative correction graphs that do contain loops . for constant @xmath114 and @xmath102 the effective potential
is related to the source via @xmath117 , so that @xmath102 does indeed break any symmetry that @xmath118 might possess .
the significance of @xmath118 is that when @xmath102 is zero , we can write @xmath118 as @xmath119 where @xmath17 and @xmath15 are spontaneously broken and normal vacua in which @xmath120 is nonzero and @xmath121 is zero . in the analyses of classical potentials such as @xmath122 and classical kinetic energies
such as @xmath123 presented above , the classical field @xmath9 represented @xmath114 , the potential @xmath124 represented @xmath118 , the kinetic energy represented @xmath125 , and in the @xmath116 green s functions only tree approximation graphs were included .
in this way the search for non - trivial minima of @xmath124 is actually a search for states @xmath17 in which @xmath126 would be negative .
thus while the analyses presented above in sec .
iv looked to be classical they actually had a quantum - mechanical underpinning with the classical field being a c - number vacuum matrix element of a q - number quantum field .
it is in this way that the classical analyses presented above are to be understood . to understand the nature of a broken symmetry vacuum it is instructive to reconsider @xmath101 .
it is associated with a system @xmath93 to which an external field has been added .
this external field breaks the symmetry by hand at the level of the lagrangian since an effective potential such as @xmath127 would be lopsided with one of its minima lower than the other , and would not have any @xmath128 symmetry .
such a situation is analogous to that found in a ferromagnet . in the presence of an external magnetic field ( cf .
@xmath102 ) all the spins line up in the direction of the magnetic field .
if one is above the critical temperature , then if one removes the magnetic field the spins flop back into a configuration in which the net magnetization is zero .
however , if one is below the critical point , the spins stay aligned and remember the direction of the magnetic field after it has been removed ( hysteresis ) .
moreover , if the magnetic field is taken to point in some other direction , below the critical point the spins will remember that direction instead , and will remain aligned in that particular direction after the magnetic field is removed . for a spherically symmetric ferromagnetic system below the critical point one can thus align the magnetization at any angle @xmath81 over a full @xmath129 to @xmath130 range , and have it remain aligned after the magnetic field is removed .
given all the different orientations of the magnetization that are possible below the critical point , we need to determine in what way we can distinguish them . so consider a single spin pointing in the @xmath131-direction with spin up , and a second spin pointing at an angle @xmath81 corresponding to a rotation through an angle @xmath81 around the @xmath132 axis .
for these two states the overlap is given by @xmath133 with the overlap being nonzero and with the two states thus necessarily being in the same hilbert space .
suppose we now take an @xmath134-dimensional ensemble of these same sets of spin states and evaluate the overlap of the state with all spins pointing in the z - direction and the state with all spins pointing at an angle @xmath81 .
this gives the overlap @xmath135 in the limit in which @xmath134 goes to infinity this overlap goes to zero . with this also being true of excitations built out of these states ,
the two states are now in different hilbert spaces .
thus broken symmetry corresponds to the existence of different , inequivalent vacua , and even though the various vacua all have the same energy ( i.e. degenerate vacua ) , the vacua are all in different hilbert spaces
. the quantum hilbert spaces associated with the various minima of the effective potential ( i.e. the differing states @xmath136 in which the vacuum expectation values @xmath137 are evaluated ) become distinct in the limit of an infinite number of degrees of freedom , even though they would not be distinct should @xmath134 be finite .
broken symmetry is thus not only intrinsically quantum - mechanical , it is intrinsically a many - body effect associated with an infinite number of degrees of freedom . whether or not a hamiltonian @xmath93 possesses such a set of degenerate vacua is a property of @xmath93 itself .
it is not a property of the external field @xmath102 .
the role of the external field is solely to pick one of the vacua , so that the system will then remain in that particular vacuum after the external field is removed . whether or not the system is actually able to remember the direction of the external field after it has been removed is a property of the system itself and not of the external field . to underscore the need for an infinite number of degrees of freedom ,
consider a system with a finite number of degrees of freedom such as the one - dimensional , one - body , quantum - mechanical system with potential @xmath138 and hamiltonian @xmath139 .
like the field - theoretic @xmath122 , the potential @xmath140 has a double - well structure , with minima at @xmath141 .
however the eigenstates of the hamiltonian can not be localized around either of these two minima .
rather , since the hamiltonian is symmetric under @xmath142 , its eigenstates can only be even functions or odd functions of @xmath143 , and must thus must take support in both of the two wells .
wave functions localized to either of the two wells are in the same hilbert space , as are then linear superpositions of them , with it being the linear combinations that are the eigenstates . thus with a finite number of degrees of freedom , wave functions localized around the two minima are in the same hilbert space .
it is only with an infinite number of degrees of freedom that one could get inequivalent hilbert spaces . now if the role of @xmath102 is only to pick a vacuum and not to make the chosen state actually be a vacuum , we need to inquire what is there about the @xmath102 dependence of the theory that might tell us whether or not we do finish up in a degenerate vacuum when we let @xmath102 go to zero .
the answer to this question is contained in @xmath101 , with @xmath101 needing to be a multiple - valued function of @xmath102 , with @xmath144 vanishing on one branch of @xmath101 in the limit in which @xmath102 goes to zero , while not vanishing on some other one . as a complex function of @xmath102 the function
@xmath101 has to have one or more branch points in the complex @xmath102 plane , and thus have some inequivalent determinations as @xmath102 goes to zero .
these different determinations correspond to different phases , with it being the existence of such inequivalent determinations that is the hallmark of phase transitions . to appreciate the point consider the two - dimensional ising model of a ferromagnet in the presence of an external magnetic field @xmath145 at temperature @xmath146 .
in the mean - field approximation the free energy per particle is given by @xmath147 , \label{m22}\end{aligned}\ ] ] where @xmath13 is the critical temperature and @xmath10 is the magnetization .
at the minimum where @xmath148 the magnetization obeys @xmath149 when @xmath146 is greater than @xmath13 the magnetization can only be nonzero if @xmath145 is nonzero .
however , if @xmath146 is less than @xmath13 one can have a nonzero @xmath10 even if @xmath145 is zero , and not only that , for every non - trivial @xmath10 there is another solution with @xmath150 .
since @xmath151 is an even function of @xmath10 , solutions of either sign for @xmath10 have the same free energy .
symmetry breaking is thus associated with a degenerate vacuum energy .
if we take @xmath145 to be complex , set @xmath152 , and set @xmath153 , @xmath154 , then when @xmath145 is nonzero we can set @xmath155 . with the logarithm term in the free energy having branch points in the complex @xmath145 plane whenever @xmath156 , we see that branch points occur when @xmath157 , @xmath158 .
thus as required , the free energy is a multiple - valued - function in the complex @xmath145 plane , with branch points on the imaginary @xmath145 axis .
while @xmath159 only has two real solutions for any given @xmath160 , it has an infinite number of pure imaginary solutions , and these are reflected in the locations of the branch points of @xmath161 .
a second example of multiple - valuedness may be found in the double - well potential @xmath122 given in sec .
iva in the presence of a constant source @xmath102 .
solutions to the theory are constrained to obey @xmath162 and are of the form @xmath163^ * , \label{m25}\end{aligned}\ ] ] where @xmath164 if we set @xmath165 , then when @xmath97 , the solutions are given by @xmath166 , @xmath167 , @xmath168 , just as found in sec iva .
however , suppose instead we fix @xmath169 , and treat @xmath9 as a multiple - valued function of @xmath102 .
then , because of the cube root in the @xmath170 term , as we set @xmath102 to zero we obtain three determinations of @xmath171 , viz .
@xmath172 , @xmath173 , @xmath174 . with @xmath97 ,
these determinations then precisely give the previous @xmath166 , @xmath167 , @xmath168 solutions . with this multiple - valuedness then propagating to @xmath124 and @xmath175 when they are evaluated in these three solutions ,
i.e. when we set @xmath176 , @xmath177 and obtain @xmath178-\frac{\mu^2}{4}\left[i^{2/3}(p+iq)^{2/3}+{\rm c.~c.}\right]-\frac{\mu^4}{6\lambda^2 } , \label{m27}\end{aligned}\ ] ] we see that in any solution @xmath101 is indeed a multiple - valued function of @xmath102 , and see that from any one solution we can derive the others by analytic continuation , with the limit @xmath179 having multiple determinations .
while we need to use the effective potential @xmath118 to explore symmetry breaking by an elementary scalar field , it turns out that even though @xmath96 was only introduced as an intermediate step in order to get to @xmath109 , for symmetry breaking by fermion @xmath180 bilinears we need to use @xmath181 directly , where @xmath182|\omega_m\rangle .
\label{m28}\end{aligned}\ ] ] specifically , consider the chiral symmetric nambu - jona - lasinio ( njl ) action with no fermion mass term of the form @xmath183
^ 2-\frac{g}{2}[\bar{\psi}i\gamma_5\psi]^2\right ] .
\label{m29}\ ] ] as such it is a relativistic generalization of the bcs model . in the mean field , hartree - fock approximation one introduces a trial wave function parameter @xmath184 that is not in the original action , and then decomposes the action into two pieces , a mean - field piece and a residual - interaction piece according to : @xmath185 + \int d^4x \left[-\frac{g}{2}\left(\bar{\psi}\psi-\frac{m}{g}\right)^2-\frac{g}{2}\left(\bar{\psi}i\gamma_5\psi\right)^2\right ] .
\label{m30}\end{aligned}\ ] ] one then tries to show that in the mean - field sector nonzero @xmath184 is energetically favored . if we define @xmath186 and define @xmath187 , then in the mean - field sector we need to show that the @xmath188 associated with @xmath189 is less than @xmath190 and thus energetically favored . now previously we had identified @xmath188 as @xmath191 , where @xmath192 .
if the state @xmath193 is one in which @xmath194 is zero , we can write @xmath188 as @xmath195 to thus enable us to compare two candidate vacua for @xmath196 , with a view to determining whether a vacuum with nonzero @xmath197 is energetically favored . in terms of green s functions @xmath188
can be evaluated via @xmath198 where the @xmath199 are the connected @xmath180 green s functions as evaluated in the theory in which @xmath200 , viz . that associated with @xmath93 . unlike the elementary scalar field @xmath118
, @xmath188 has no tree approximation contribution and is entirely generated by radiative loops .
thus while we need to use @xmath118 to explore symmetry breaking by an elementary scalar field with @xmath118 being the energy density difference between different candidate vacua of @xmath201 , for symmetry breaking by a fermion composite we use @xmath188 instead , with @xmath188 being the energy density difference between different candidate vacua of @xmath202 .
we shall explore this issue in more detail below , while noting now that in distinguishing between an elementary ( i.e. god given " ) higgs boson and a composite one there are even differences in setting up the formalism in the first place .
given that the existence of the higgs boson has now been confirmed , we need to ask what exactly the field @xmath9 represents .
there are two possibilities .
it is either a q - number elementary field @xmath9 that appears in the fundamental @xmath28 lagrangian ( to thereby be on an equal footing with the fundamental quarks , leptons and gauge bosons ) , or it is generated as a dynamical bound state , with the field in a dynamically induced higgs potential then being the c - number matrix element @xmath203 , a dynamical bilinear fermion condensate .
the mexican hat potential is thus either part of the fundamental lagrangian or it is generated by dynamics .
if the higgs field is elementary , then while the potential @xmath122 would be its full quantum - mechanical potential , the discussion given earlier of the minima of the potential would correspond to a c - number tree approximation analysis with the @xmath9 that appeared there being the c - number @xmath204 .
however , in the fermion condensate case there is no tree approximation , with the theory being given by radiative loop diagrams alone . to see how to generate the mexican hat potential in this case we consider the nambu - jona - lasinio ( njl ) four - fermion model . via an infinite summation of massless graphs with zero - momentum point @xmath213 insertions.,width=316,height=124 ] given this gap equation , for @xmath214
we can calculate the one loop mean - field vacuum energy density @xmath188 as a function of a constant @xmath184 via the graphs of fig .
this yields : @xmath215-\frac{m^2}{2 g } \nonumber\\ & = & \frac{m^4}{16\pi^2}{\rm
ln}\left(\frac{\lambda^2}{m^2}\right)-\frac{m^2m^2}{8\pi^2}{\rm ln}\left(\frac{\lambda^2}{m^2}\right)+\frac{m^4}{32\pi^2}. \label{m36}\end{aligned}\ ] ] while the energy @xmath216 $ ] has quartic , quadratic and logarithmically divergent pieces , the subtraction of the massless vacuum energy @xmath217 $ ] removes the quartic divergence , with the subtraction of the self - consistent induced mean field term @xmath218 then leaving @xmath188 only logarithmically divergent .
we recognize the resulting logarithmically divergent @xmath188 as having a local maximum at @xmath200 , and a global minimum at @xmath219 where @xmath10 itself is finite .
we thus induce none other than a dynamical double - well mexican hat potential , and identify @xmath10 as the matrix element of a fermion bilinear according to @xmath220 .
if instead of looking at matrix elements in the translationally - invariant vacuum @xmath17 we instead look at matrix elements in coherent states @xmath22 where @xmath221 is now spacetime dependent , we then find @xcite that the resulting mean field effective action has a logarithmically divergent part of the form @xmath222 . \label{m37}\end{aligned}\ ] ] here the kinetic energy term is the analog of the @xmath223 term given earlier in the expansion of @xmath96 around the point where all momenta vanish . in terms of the quantity @xmath224 to be introduced below , @xmath225 is given by @xmath226 .
if we introduce a coupling @xmath227 to an axial gauge field @xmath228 , on setting @xmath229 the effective action becomes @xmath230}. \nonumber\\ \label{m38}\end{aligned}\ ] ] we recognize this action as a double - well ginzburg - landau type higgs lagrangian , only now generated dynamically .
we thus generalize to the relativistic chiral case gorkov s derivation of the ginzburg - landau order parameter action starting from the bcs four - fermion theory . in the @xmath231 effective action associated with the njl model
there is a double - well higgs potential , but since @xmath221 is a c - number , @xmath232 does not itself represent a q - number scalar field .
and not only that , unlike in the elementary higgs case , the second derivative of @xmath233 at the minimum where @xmath219 is not the mass of a q - number higgs boson .
rather , as we now show , the q - number fields are to be found as collective modes generated by the residual interaction , and it is the residual interaction that will fix their masses . moreover , as we will see in sec .
vii , when we dress the point njl vertices in figs .
( 2 ) and ( 3 ) , the dynamical higgs boson will move above the threshold in the fermion - antifermion scattering amplitude , become unstable , and acquire a width .
this width has the potential to discriminate between an elementary higgs boson and a dynamical one . to find the collective modes we calculate the scalar and pseudoscalar sector green s functions @xmath234 , @xmath235 , as is appropriate to a chiral - invariant theory .
if first we take the fermion to be massless ( i.e. setting @xmath236 where @xmath237 ) , to one loop order in the four - fermion residual interaction we obtain @xmath238 the scattering matrices in the two channels are given by @xmath239 and with @xmath240 given by the gap equation , near @xmath241 both scattering matrices behave as @xmath242 to give degenerate ( i.e. chirally symmetric ) scalar and pseudoscalar tachyons at @xmath241 ( just like fluctuating around the local maximum in a double - well potential ) , with @xmath15 thus being unstable . developed as an infinite summation of massless graphs , each with two point @xmath213 insertions carrying momentum @xmath243 ( shown as external lines ) , with all other point @xmath213 insertions carrying zero momentum.,width=326,height=76 ] however ,
suppose we now take the fermion to have nonzero mass @xmath10 ( i.e. we set @xmath244 ) .
as per the summation for @xmath245 given in fig .
( 3 ) and its @xmath246 analog , we obtain @xmath247 given the form for @xmath240 , we find a dynamical pseudoscalar goldstone boson bound state at @xmath248 and a dynamical scalar higgs boson bound state at @xmath249 ( @xmath250 ) .
the two dynamical bound states are not degenerate in mass ( spontaneously broken chiral symmetry ) , and the dynamical higgs scalar mass @xmath251 is twice the induced mass of the fermion .
the most crucial question for the higgs boson is determining its underlying nature .
thus we need to find some experimental way to determine whether the higgs boson is elementary or dynamical .
if the higgs field is elementary , one would have to treat the higgs lagrangian as a bona fide quantum field theory with an a priori double - well potential being present in the fundamental lagrangian itself .
the great appeal of a higgs potential of the specific form @xmath122 is that the radiative corrections that it generates are renormalizable .
thus not only would the massive gauge boson sector of the theory be renormalizable ( because of the higgs mechanism ) , the higgs sector itself would be too .
moreover , with an elementary higgs field electroweak radiative corrections are not only finite , they are straightforwardly calculable .
however , having an elementary higgs field also has some disturbing consequences .
first we note that the radiative corrections in the higgs sector itself lead to a quadratically divergent higgs self energy .
while this divergence can be made finite by renormalization , there is no control on the ensuing magnitude that the higgs mass would take , as it would be as large as the regulator masses .
these masses could be at the @xmath252 gev or so grandunified scale ( a very large scale if the proton lifetime is to be greater than its current experimental lower bound ) or even at the @xmath253 gev quantum gravitational planck scale .
such mass values are orders of magnitude larger than the higgs boson s observed 125 gev mass .
in order to resolve such a disparity ( known as the hierarchy problem ) it had long been conjectured that there would be a supersymmetry between bosons and fermions , and with fermion and boson loops having opposite signs , a fermionic superparticle could then cancel the quadratic divergence in the higgs self - energy . for it to do so to the degree needed , the fermionic superparticle would need to be close in mass to the higgs boson itself .
however , data from the very same lhc that was used to find the higgs boson have sharply curtailed the possibility that there might be any fermionic superparticle in the requisite mass region . thus the higgs self - energy problem is open at the present time .
a second concern for an elementary higgs field is that the group theoretic structure of the @xmath5 theory does not at all constrain the yukawa couplings of the higgs field to the fundamental fermions .
thus while the higgs field can give masses to the quarks and leptons , those masses are totally unconstrained , with all the yukawa coupling constants being free parameters that one has to introduce by hand .
while embedding @xmath5 in some grandunified theory of the strong , electromagnetic and weak interactions , say , might solve this problem , the issue is open at the present time .
a third concern for an elementary higgs field is that , as had been noted earlier , the very minimization of the higgs potential generates an enormous contribution to the cosmological constant . for a one tev or
so higgs mass breaking scale one would get a cosmological constant that would be of order @xmath54 times larger than the standard einstein gravitational theory could possibly permit .
this problem is very severe , and it also is open at the present time .
a fourth concern for an elementary higgs field is an aesthetic , in principle one . at the level of the @xmath28 lagrangian all fermions and gauge bosons are massless and all coupling constants are dimensionless . the only place where there is a fundamental mass scale is in the @xmath254 term in the higgs potential
. it would be much more natural and elegant if this @xmath255 scale were to be generated dynamically in a then scale invariant theory that would then possess no fundamental scale at all .
it is thus of interest to note that all of these concerns can be addressed if the higgs boson is dynamical .
moreover , there already are two working models that we know of in which the symmetry breaking actually is done dynamically , the bcs theory , and the generation of a goldstone boson pion in qcd . in the bcs theory the symmetry breaking
is done by a dynamically generated nonzero fermion bilinear vacuum expectation value @xmath256 in a theory that only contains electrons and phonons while not containing any elementary scalar fields whatsoever . for the pion
, we note that the qcd local color @xmath257 lagrangian of quarks and gluons possesses a global chiral flavor symmetry .
this chiral symmetry is broken dynamically by qcd dynamics to yield a goldstone pion .
this pion then acquires a mass via the weak interaction since the electroweak @xmath5 action breaks the flavor chiral symmetry intrinsically at the level of weak interaction lagrangian itself , to thereby make the chiral favor symmetry be only an approximate one .
the generation of a goldstone pion in qcd is particularly of interest since unlike the njl model , in the qcd case it occurs in a theory that is renormalizable , and in which all coupling constants are dimensionless and all quark and gluon masses are zero at the level of the lagrangian . in such a theory not only is there no fundamental @xmath254 term , there is not even any elementary @xmath9 at all , and the theory is scale invariant at the level of the lagrangian .
so let us suppose that dynamical symmetry breaking occurs in some renormalizable yang - mills theory of interacting massless fermions and gauge bosons .
in such a case , if nonzero gauge boson masses are produced by a dynamical higgs mechanism , renormalizability would not be impaired since bound state production in a renormalizable theory does not violate renormalizability . without needing to specify any particular such yang - mills theory ( i.e. without needing to specify the group structure or the specific matter content of the theory ) , we note that if the symmetry breaking occurs at all , then since all such theories have no divergences higher than logarithmic , there will be no quadratic divergence associated with any dynamical scalar bound states that might be produced .
thus with a dynamical higgs boson there is no higgs self - energy hierarchy problem at all .
as regards the couplings of fermions to the higgs boson , these couplings are given as the residues of the dynamical poles in the requisite channels .
thus they are determined by the theory itself and are not free parameters at all . to see how things work ,
let us consider the njl model as a stand - in for a renormalizable field theory . in its @xmath258 and @xmath259 channels
there are scalar and pseudoscalar bound states , and near the respective poles the scattering amplitudes behave as : @xmath260 we thus identify the yukawa coupling of the scalar and psuedoscalar bound states to a fermion anti - fermion pair to be @xmath261 and @xmath262 , where the values of @xmath263 and @xmath264 are not fixed by hand but by the theory itself , with the equality of @xmath263 and @xmath264 that is found reflecting the underlying chiral symmetry of the njl theory .
as regards the cosmological constant problem , we had noted above that the self - consistent mean - field treatment of the njl model automatically generated an @xmath218 term in the mean - field action , with this term serving to cancel the quadratic divergence in the vacuum energy .
unlike the situation with an elementary higgs field where there is no control of the vacuum energy , with dynamical symmetry breaking we see that contributions to the vacuum energy are potentially controllable . breaking symmetries dynamically thus provides a good starting point to address the cosmological constant problem . via an infinite summation of massless graphs with
zero - momentum dressed @xmath213 insertions.,width=316,height=124 ] developed as an infinite summation of massless graphs , each with two dressed @xmath213 insertions carrying momentum @xmath243 ( shown as external lines ) , with all other dressed @xmath213 insertions carrying zero momentum.,width=326,height=76 ] a discussion of how things might work in the renormalizable situation can be found in the conformal gravity and conformal symmetry studies given in @xcite and @xcite . in these studies conformal symmetry is used to address the cosmological constant problem with the cosmological constant being zero if the conformal symmetry is unbroken , while being constrained in a phenomenologically acceptable way if the conformal symmetry is broken dynamically .
in addition , the conformal symmetry is realized via scaling with anomalous dimensions , with the green s function associated with the insertion of a soft @xmath180 into the inverse fermion propagator being found to behave as @xmath265 .
with such a form the dynamical dimension of @xmath180 is reduced from three to two , and the dynamical dimension of @xmath266 ^ 2 $ ] is reduced from six to a renormalizable four .
the four - fermion theory is thus made renormalizable by dressing its point vertices in this particular way , with the point - coupled vertices of the njl model with @xmath267 being replaced by dressed vertices with @xmath265 . with the @xmath265 dressing all infinities in the particle physics sector are now under control , and with conformal gravity being a renormalizable ( and unitary @xcite ) theory , all infinities in both particle physics and gravity ( and thus the infinite negative energy sea contribution to the cosmological constant ) are now under control . for our purposes here we note that once one dresses the vertices as exhibited in figs .
( 4 ) and ( 5 ) , quantities such @xmath268 will develop a non - trivial momentum dependence , so that @xmath268 will then have a discontinuity in the complex @xmath269 plane . in consequence of this
, the pole in @xmath270 will no longer be at the particle - antiparticle threshold but will instead be above threshold and not be on the real @xmath269 axis @xcite .
the dynamically induced higgs boson must thus be an unstable above - threshold resonance and have a nonzero width .
it is in this respect that a composite higgs boson will most differ from an elementary one , with the width of the higgs boson thus being a diagnostic that could potentially distinguish an elementary higgs boson from a dynamical one .
however , if the higgs boson is to be dynamical , the dynamical theory in which it is to be generated would then have to be able to reproduce those aspects of the radiative correction structure associated with an elementary higgs field theory that have been tested .
to this end we note that following gaussian integration on a dummy scalar field variable @xmath271 and a dummy vector field variable @xmath272 , the path integral representation of the generator @xmath273 of fermion green s functions in the njl theory can be written as : @xmath274[d\psi]\exp\left[i\int d^4x \left(i\bar{\psi}\gamma^{\mu}\partial_{\mu}\psi-\frac{g}{2}(\bar{\psi}\psi)^2+\bar{\eta}\psi+\bar{\psi}\eta\right)\right ] \nonumber\\ & = & \int [ d\bar{\psi}][d\psi ] [ d\sigma][dv_{\mu}]\exp\bigg{[}i\int d^4x \bigg{(}i\bar{\psi}\gamma^{\mu}\partial_{\mu}\psi-\frac{g}{2}(\bar{\psi}\psi)^2+\frac{g}{2}\left(\frac{\sigma}{g}-\bar{\psi}\psi\right)^2 \nonumber\\ & + & \frac{1}{2}(h v_{\mu}-\partial_{\mu}\sigma)(h v^{\mu}-\partial^{\mu}\sigma)+\bar{\eta}\psi+\bar{\psi}\eta\bigg{)}\bigg { ] } \nonumber\\ & = & \int [ d\bar{\psi}][d\psi ] [ d\sigma][dv_{\mu}]\exp\bigg{[}i\int d^4x \bigg{(}i\bar{\psi}\gamma^{\mu}\partial_{\mu}\psi-\sigma\bar{\psi}\psi + \frac{\sigma^2}{2g}+ \nonumber\\ & + & \frac{1}{2}(h^2 v_{\mu}v^{\mu}-h\partial_{\mu}\sigma v^{\mu}-hv_{\mu}\partial^{\mu}\sigma + \partial_{\mu}\sigma\partial^{\mu}\sigma)+\bar{\eta}\psi+\bar{\psi}\eta\bigg{)}\bigg { ] } , \label{m44}\end{aligned}\ ] ] where @xmath275 is a constant . on setting @xmath275 to zero ,
the integrand becomes independent of @xmath272 , the @xmath272 path integration becomes irrelevant , and @xmath276 reduces to @xmath274[d\psi ] [ d\sigma]\exp\bigg{[}i\int d^4x \bigg{(}i\bar{\psi}\gamma^{\mu}\partial_{\mu}\psi-\sigma\bar{\psi}\psi + \frac{\sigma^2}{2g}+ \frac{1}{2}\partial_{\mu}\sigma\partial^{\mu}\sigma+\bar{\eta}\psi+\bar{\psi}\eta\bigg{)}\bigg{]}. \nonumber\\ \label{m45}\end{aligned}\ ] ] as we see , the fermion green s functions of the njl theory are given as the fermion green s functions of a scalar field theory whose action is precisely the njl mean field action as evaluated in coherent states in which a @xmath271 kinetic energy term is generated @xcite . in consequence ,
the perturbative expansions in the two theories are in one to one correspondence . however , in this scalar field theory there is no source term @xmath277 for the scalar field ( in a true fundamental higgs field lagrangian there would be such a source term ) , and thus the scalar field theory only generates green s functions with external fermion legs and does not generate any green s functions with external scalar field legs .
thus in the dynamical higgs case one can generate the fermion green s functions using an elementary higgs field theory in which the elementary higgs field only role is to contribute internally in feynman diagrams and never to appear in any external legs .
in such a case , the all - order iteration of internal @xmath271 exchange diagrams then generates the dynamical higgs and goldstone poles in @xmath270 and @xmath278 .
moreover , as we had noted above , the dynamically generated higgs boson will be an above - threshold resonance and have a nonzero width . since a fundamental higgs boson can contribute both off shell and on shell while the dummy field @xmath271 can only contribute off shell ,
it is in the on - shell behavior in the higgs sector and through the width that it would acquire that one may ultimately be able to distinguish between an elementary higgs boson and a dynamical one .
with the vacuum of quantum field theory being a dynamical one , in a sense einstein s ether has reemerged . only it has reemerged not as the mechanical ether of classical physics that was excluded by the michelson - morley experiment , but as a dynamical , quantum - field - theoretic one full of dirac s negative energy particles , an infinite number of such particles whose dynamics can spontaneously break symmetries .
the type of physics that would be taking place in this vacuum depends on how symmetries are broken , i.e. on whether the breaking is by elementary higgs fields or by dynamical ones .
if the symmetry is broken by an elementary higgs field , then the higgs boson gives mass to fundamental gauge bosons and fermions alike .
however , if the breaking is done dynamically , then it is the structure of an ordered vacuum itself that generates masses , with the mass generation mechanism in turn then producing the higgs boson . in the dynamical case
then mass produces higgs rather than higgs produces mass . in the dynamical case we should not be thinking of the higgs boson as being the god particle " .
rather , if anything , we should be thinking of the vacuum as being the god vacuum " .
this paper is based on a physics department colloquium that the author gave at the university of connecticut in november 2013 , and a presentation that the author gave to physics majors at the university of connecticut in march 2014 . for the reader who wishes to delve further into the subject , a nice discussion of the history and development of the higgs mechanism
may be found in f. close , the infinity puzzle : the personalities , politics , and extraordinary science behind the higgs boson , " ( oxford university press , oxford uk , 2013 ) . | in 2013 the nobel prize in physics was awarded to francois englert and peter higgs for their development in 1964 of the mass generation mechanism ( the higgs mechanism ) in local gauge theories .
this mechanism requires the existence of a massive scalar particle , the higgs boson , and in 2012 the higgs boson was finally observed at the large hadron collider after being sought for almost half a century . in this pedagogical article
we review the work of these nobel recipients and discuss its implications .
we approach the topic from the perspective of a dynamically generated higgs boson that is a fermion - antifermion bound state rather than an elementary field that appears in an input lagrangian .
in particular , we emphasize the connection with the bcs theory of superconductivity .
we identify the double - well higgs potential not as a fundamental potential but as a mean - field effective lagrangian with a dynamical higgs boson being generated through a residual interaction that accompanies the mean - field lagrangian .
we discuss what we believe to be the key challenge raised by the discovery of the higgs boson , namely determining whether it is elementary or composite , and suggest that the width of the higgs boson might serve as a suitable diagnostic for discriminating between an elementary higgs boson and a composite one . |
the inflation paradigm of the early universe provides a compelling explanation for many big puzzles in the standard big bang cosmology , such as the problems of homogeneity , isotropy and flatness of the universe @xcite .
this period of accelerated expansion in the early universe predicts a nearly scale - invariant perturbation power spectrum , which has already been supported by the measurement of temperature fluctuations in the cosmic microwave background radiation @xcite . in spite of many phenomenological successes of the inflation based on effective field theory ,
there exist serious problems , concerning the origin of the scalar field driving inflation , namely the singularity problem and the trans - plankian problem .
therefore , it is expected that inflation should be realized in a more natural way from some fundamental theory of microscopic physics .
string theory is one of the most promising candidates for the fundamental microscopic physics , so it is very important to try to embed the inflation scenarios into string theory .
one possible inflation scenario within the framework of string theory is the brane inflation , in which the inflation is driven by the potential between the parallel dynamic brane and anti - brane @xcite .
the distance between the branes in the extra compactified dimensions plays the role of the inflaton field .
however , generically , getting a sufficiently flat inflaton potential in brane inflation models is not an easy thing @xcite .
a rather realistic slow roll inflation scenario based on string theory was first proposed by kachru et al .
@xcite . in @xcite , by introducing some @xmath13-branes in a warped geometry in type iib superstring theory to break supersymmetry , the authors successfully lift the ads vacuum to a metastable de sitter vacuum with sufficiently long lifetime .
this mechanism is often called kklt mechanism . furthermore , by putting an extra pair of the brane and anti - brane in this scenario , one can successfully realize a slow roll inflation model , namely the kklmmt model @xcite . in this model , the anti - brane is fixed at the bottom of a warped throat , while the brane is mobile and experiences a small attractive force towards the anti - brane .
the inflation ends when the brane and the anti - brane collide and annihilate , initiating the hot big bang epoch .
the annihilation of the brane and anti - brane makes the universe settle down to the string vacuum state that describes our universe . during the process of the brane collision , cosmic strings
are copiously produced @xcite . for a well - worked inflationary scenario ,
the production of topological defects other than cosmic strings must be suppressed by many orders of magnitude . for extensive studies on the kklmmt model and other types of brane inflation models ,
see , e.g. , @xcite .
recently , the wmap team released the five - year data @xcite . for the @xmath2cdm model ,
wmap five - year data show that the index of power spectrum satisfies . ]
@xmath3 combining wmap with sdss and snia , the result is @xmath4 which is a little bluer than the wmap three - year result , though the red power spectrum is still favored at the level of @xmath5 cl .
meanwhile , the running of the spectral index is not favored anymore . with wmap five - year data
only , the running of the spectral index is . ]
@xmath6 combining with sdss and snia data , the result is . ]
@xmath7 it is notable that the result of the spectral index in five - year wmap is slightly larger than that in three - year wmap .
more evidently , the result of the running of the spectral index in five - year wmap is much smaller than that in three - year wmap .
finally , let s see the result of the amplitude of the primordial gravitational waves .
when combining wmap data with sdss and snia data , the tensor - to - scalar ratio @xmath8 is limited under a much tighter bound : . ]
@xmath9 based on the wmap three - year results , huang et al .
@xcite investigated brane inflation models and showed that the kklmmt model can not fit wmap+sdss data at the level of one standard deviation and a fine - tuning ( at least eight parts in a thousand ) is needed at the level of two standard deviations .
now , the wmap data are updated to the five - year ones , so it is meaningful to see how the status of brane inflation is affected by the arrival of the new wmap data .
therefore , in this paper , we address this issue , i.e. , we revisit brane inflation with the wmap five - year results .
this paper is organized as follows : in section [ sec : dp ] , we discuss a simple brane inflation model neglecting the problem of dynamic stabilization . in section [ sec : kklmmt ] , we focus on the kklmmt model and investigate to what degree
the model is fine - tuned under the wmap five - year data .
the conclusion is given in the last section .
first , let s consider some general brane inflation models . consider a pair of @xmath10 and @xmath11-branes ( @xmath12 ) filling the four large dimensions and separated from each other in the extra six dimensions that are compactified .
the inflaton potential is given by @xcite @xmath13 where @xmath14 is an effective cosmological constant on the brane provided by brane tension , and the second term in ( [ potential1 ] ) comes from the attractive force between the branes .
the parameter @xmath15 satisfies the relation @xmath16 , where @xmath17 is the number of transverse dimensions .
obviously , we have @xmath18 due to @xmath19 .
corresponding to the @xmath20-folding number @xmath21 before the end of inflation , the inflaton field @xmath22 takes the value @xmath23^{1/(n+2 ) } , \label{begin1}\ ] ] where @xmath24 is the reduced planck mass of four dimensional world .
the slow - roll parameters can be given as @xmath25@xmath26@xmath27due to the fact that @xmath28 is much less than @xmath24 , @xmath29 is negligible .
hence , the amplitude of the tensor perturbations is negligible @xcite .
one then obtains the spectral index and its running : @xmath30@xmath31it is easy to see that the running of the spectral index @xmath32 is also negligible .
, the region with @xmath33 is not allowed ( yellow region ) . left : wmap five - year results only ; right : wmap five - year data combined with snia and bao data .
, width=316 ] figure [ ns1 ] shows the spectral index for different parameter @xmath15 and @xmath20-folding number @xmath21 .
the region of @xmath33 is not allowed by theory , since the number of transverse dimensions is less than six .
we focus on the two cases of @xmath34 and 4 corresponding respectively to the @xmath35 and @xmath36-brane cases . from fig .
[ ns1 ] , we see that the brane inflation model is consistent with the wmap data within some range of @xmath21 . on the other hand ,
it has been proposed in @xcite that the @xmath20-folding number relevant to observations should be @xmath37 so , we shall check whether the fitting of wmap five - year data can give a reasonable @xmath20-folding number consistent with the result of ( [ e - folds ] ) , namely , @xmath38 . .fitting results for the number of @xmath20-folds @xmath21 [ cols="^,^,^,^,^",options="header " , ] table [ tab1 ]
is the result of the fitting to the wmap data ( combined with other observational data ) for the number of @xmath20-folds @xmath21 .
it is clear to see that the fitting values of the @xmath20-folding number are generally consistent with the result of @xcite , @xmath39 , within @xmath40 range .
therefore , testing with the wmap five - year data , one finds that the model survives at the level of 1@xmath41 .
furthermore , let us consider the running of the spectral index .
we plot @xmath32 vs. @xmath21 in fig .
[ as1 ] . from the left panel of this figure
, we find that though the brane inflation model can not yield the right results of the running of the spectral index given by wmap five - year data in 1@xmath41 range with a reasonable @xmath20-folding number , but it can be consistent with the observational results in 2@xmath41 range . from the right panel of fig .
[ as1 ] , one finds that , for wmap5+sn+bao , the model is even consistent with the data in 1@xmath41 range .
[ model1 ] shows the joint two - dimensional marginalized distribution of the spectral index @xmath42 and its running @xmath32 , for wmap five - year data . from this figure
, we see that the trajectories of the brane inflation model ( d5 and d3-brane cases ) are within the region of 1@xmath41 cl . is not allowed ( yellow region ) . left : wmap five - year results only ; right : wmap five - year data combined with snia and bao data .
, width=316 ] by modifying the camb code @xcite , we also compare the angular power spectrum of the model with the wmap five - year data , see fig .
[ angular ] . in the left panel
, we see that the angular power spectrum produced by the model is generally consistent with the wmap five - year data , although the cases with @xmath34 and @xmath43 have little greater amplitudes of the low multipole moments ( basically quadrupole and octopole ) compared with the best fit of the wmap five - year data
. however , due to the cosmic variance and statistical errors , the low multipole moments have large error bars , the model prediction is consistent with the observational data .
the above analyses show that the general brane inflation model can survive confronting the wmap five - year data . when considering the wmap five - year result of the spectral index , the model survives at the level of 1@xmath41 .
this conclusion is different from that of the analysis of wmap three year data .
it was indicated in @xcite that , with the wmap three year data , the brane inflation model for @xmath43 is near the boundary of the wmap3 only and can not fit the wmap3+sdss data at the level of 1@xmath41 ; the case with @xmath34 is outside the range allowed by wmap3 only or wmap3+sdss at the level of 1@xmath41 .
so , we find that the wmap five - year data save the brane inflation model to the level of 1@xmath41 . however , the brane inflation model discussed above is not a realistic model .
an important problem is neglected in the above discussions , which is that the distance between the brane and the anti - brane must be larger than the size of the extra dimensional space if the inflation is slow rolling ( or @xmath44 is sufficiently small ) in this scenario .
of course , there may exist such a possibility that the potential in ( [ potential1 ] ) can emerge in other theories that we do not know currently , as pointed out in @xcite .
the first realistic brane inflation model is the so - called kklmmt model @xcite that will be discussed in the subsequent section .
the kklmmt model is set up in the type iib string theory . in this model ,
the highly warped compactifications are considered , with all moduli stabilized by the combination of fluxes and non - perturbative effects @xcite .
once a small number of @xmath13-branes are added , the vacuum can be successfully lifted to ds .
furthermore , one can add an extra pair of d3-bane and @xmath13-brane in a warped throat , with the d3-brane moving towards the @xmath13-brane that is located at the bottom of the throat .
the scenario of brane inflation can thus be realized .
the warped geometry successfully solves the @xmath44 problem , i.e. , it can provide a natural mechanism for achieving a sufficiently flat potential . in this section
, we shall discuss the kklmmt model with the wmap five - year results .
consider a type iib orientifold compactified on a calabi - yau 3-fold with fluxes , where all moduli are stabilized @xcite .
inside the bulk of the calabi - yau manifold , there are throats with warped geometry , where the metric has the approximate @xmath45 form , where @xmath46 is some orbifold of @xmath47 and the @xmath48 metric in poincare coordinates has the form @xcite @xmath49 where the @xmath50 is the warp factor @xmath51 in which @xmath52 is the radius of curvature of the @xmath48 throat .
following @xcite , we consider the potential for the kklmmt model given by @xmath53where @xmath54 is the d3-brane tension and we have the relation @xmath55 .
then we have@xmath56so the derivative of this potential is @xmath57under the slow roll approximation , we have the friedmann equation @xmath58which leads to the approximation relation @xmath59 we will use in the following calculation . the procedure to obtain the inflaton field value at the moment of the @xmath20-folding number @xmath21 before the end of inflation is as follows :
first , we get the inflaton field value at the end of inflation by using the condition @xmath60 ; then , we use the definition of @xmath20-folding number @xmath61 to obtain the field value at the moment corresponding to the e - folding number @xmath21 . therefore , we first calculate the slow - roll parameter @xmath44 : @xmath62 where @xmath63 substituting @xmath64 into the above equation , we obtain @xmath65 thus , the field value at the end of inflation @xmath66 satisfies @xmath67and we obtain @xmath68 under the slow roll approximation , we have @xmath69_{\phi _ { f}}^{\phi _ { n}}.\end{aligned}\]]substituting @xmath64 and eq .
( [ phif ] ) into the above equation , we obtain @xmath70,\end{aligned}\ ] ] or @xmath71 where @xmath72 with the value of @xmath73 , we can write @xmath74^{2 } , \label{epsilon2}\ ] ] @xmath75 @xmath76 . \label{xi2}\ ] ] the amplitude of the scalar comoving curvature fluctuations has been given in @xcite : @xmath77 thus we have @xmath78 substituting the above equation into eq .
( [ epsilon2 ] ) , we have the following expression for @xmath79 : @xmath80 the wmap five - year data give the amplitude of the primordial scalar power spectrum , @xmath81 for @xmath82 @xcite .
thus , so far , all of the slow roll parameters can be expressed as the functions of @xmath0 ( and @xmath21 ) . the spectral index and its running are @xmath83and the tensor - to - scalar ratio is @xmath84 inflation models can predict some observables , such as the spectral index , its running , and the tensor - to - scalar ratio , etc .
, from which one can link the observations to the theory . in particular
, the brane inflation models also predict the existence of cosmic superstrings that open a significant window for testing the superstring theory . in this subsection
, we shall discuss how these observables provide constraints on the parameter @xmath0 of the kklmmt model .
first , let s discuss the tensor - to - scalar ratio @xmath8 .
the tensor - to - scalar ratio is an important observable that can distinguish the inflationary cosmology from other scenarios .
also , it can be used to distinguish different inflation models because different models give distinctive predictions on this observable quantity .
although the observations still have no ability to detect the primordial gravitational waves in a convincing manner , an upper limit could be given at least .
now we try to see how this observable with such a loose upper - limit can set constraint on the kklmmt model . in fig .
[ tensor1 ] , we plot the tensor - to - scalar ratio @xmath8 vs. the @xmath20-folding number @xmath21 for the kklmmt model , according to eq .
( [ tensor ] ) . in fig .
[ tensor1 ] , we see that the larger value the parameter @xmath0 takes , the greater value the tensor - to - scalar ratio will be .
this is due to the fact that the greater the value of @xmath0 is , the faster the inflaton moves , so the larger tensor mode will be produced in the inflationary era @xcite .
currently , the tightest constraint on @xmath8 is @xmath85 ( wmap5+snia+bao ) @xcite . from fig .
[ tensor1 ] we see that the green curve ( corresponding to @xmath86 ) definitely violates this bound , and we learn that the parameter @xmath0 can not be greater than @xmath87 .
hence , we find that even such a loose bound on @xmath8 gives a so small value of @xmath0 about @xmath88 , then the true value of @xmath0 must be fine - tuned to some extent .
we hope the future cmb experiments could further constrain @xmath8 toward @xmath89 , providing true test on the kklmmt model .
the above constraint on @xmath0 can also be examined from the field value perspective .
we can calculate the change magnitude of the inflaton field from the moment when the fluctuations were generated to the end of inflation , @xmath90.\end{aligned}\ ] ] according to @xcite , for a single field inflation model , the change magnitude of the field has a relationship with the tensor - to - scale ratio as @xmath91 we plot the change magnitude of the field as a function of @xmath92 for different e - folding number @xmath21 in the left panel of fig .
[ field1 ] .
it is very explicit that if @xmath93 , the change magnitude of the field will be definitely greater than the planck mass , which might not be consistent with the conditions of effective field theory @xcite .
the right panel of fig .
[ field1 ] shows the allowed values for the parameter @xmath0 . according to eq .
( [ field and tensor ] ) , if the change of the field during the whole period of inflation is less than the planck mass , the tensor - to - scalar ratio will be surely less than @xmath89 .
thus , from right panel of fig . [ field1 ] , it can be inferred that @xmath0 should be less than @xmath94 .
therefore , to satisfy the current constraint on the tensor - to - scalar ratio and the physical bound on the change magnitude of the field value , @xmath95 , the parameter @xmath0 could not be greater than @xmath87 , which is both an observational and a physical bound on the kklmmt model . cosmic strings and other topological defects
have long been proposed as one of the candidates for the origin of structure formation @xcite , however , it has been shown that this scenario leads to predictions incompatible with observations such as the power spectrum of cosmic microwave background ( cmb ) temperature anisotropy @xcite .
nevertheless , the brane / anti - brane inflation scenario inspired from string theory naturally indicates that cosmic strings would have a small contribution to the cmb , which is compatible with the observational limits @xcite . this has led to a significant revival of all aspects of cosmic string scenario , including new theoretical motivations , phenomenological implications and direct observational searches ,
see e.g. @xcite .
the possibility of detecting the signal of the cosmic strings through astronomical observations opens a significant window to test string theory .
the current observational bound ( the upper limit ) for cosmic string tension from the cmb temperature is roughly @xcite @xmath96 and @xmath97 obviously , the kklmmt scenario predicts that the tension of these cosmic strings is very small and their contribution to cmb is well below the current limits @xcite .
however , now , we try to do this inversely , i.e. , we intend to see how this observational bound puts a constraint on the parameter @xmath0 .
this may show what the loosest observational result can tell us .
the @xmath36-brane collides with the @xmath98-brane at the end of inflation at the bottom of the throat .
the annihilation of the d3-@xmath99-branes initiates the hot big bang epoch , meanwhile d1-branes ( i.e. d - strings ) and fundamental closed strings ( i.e. f - strings ) are also produced .
the quantities of interest are the d - string tension @xmath100 and the f - string tension @xmath101 , where @xmath102 and @xmath103 are the effective tensions measured from the viewpoint of the four dimensional effective action .
since in ten dimensions , there is the relationship @xmath104 , where @xmath105 is the d3-brane tension , we have the tensions of the type iib string in the inflationary throat @xmath106 @xmath107 where @xmath108 is the string coupling .
there are also bound states of @xmath109 f - strings and @xmath110 d - strings with a cosmic string network spectrum @xcite @xmath111 in eqs .
( [ ftension ] ) and ( [ dtension ] ) both f - strings and d - strings are dependent on the string coupling @xmath108 , but the geometric mean @xmath112 is independent of @xmath108 .
thus , we can define @xcite @xmath113 obviously , for f - string we have @xmath114 , and for d - string we have @xmath115 .
the value of @xmath108 is likely in the range 0.1 to 1 @xcite .
note that @xmath116 can be converted to @xmath117 by @xmath118-duality .
then , combining with cobe normalization ( [ amplitude ] ) , we obtain the cosmic string tension in the kklmmt inflation model @xmath119 equation ( [ stension2 ] ) shows that the cosmic string tension @xmath120 depends both on the parameter @xmath0 and the number of e - folds @xmath21 .
thus , in order to see how the cosmic string tension relies on the two parameters , we plot the correlation of @xmath21-@xmath0 according to the cosmic string tension @xmath120 in fig . [ figtension ] . in fig .
[ figtension ] , the upper - right corner is the not - allowed parameter region , since it represents a too large tension and it is excluded outside the @xmath121 cl .
the smaller value the cosmic string tension takes , the deeper the curve will move toward the lower - left corner , which indicates a smaller parameter @xmath0 .
according to the @xmath122 bound ( the red curve in fig .
[ figtension ] ) , if one takes @xmath123 , the observational @xmath122 curve will give the tightest constraint on @xmath0 , namely @xmath124 ; if taking @xmath125 , the bound on the parameter @xmath0 from cosmic string tension will be looser , @xmath126 .
thus , roughly speaking , the current observational bound on the cosmic string tension tells us the following information on the parameter @xmath0 , namely @xmath127 .
the spectral index is certainly the most important observable that could distinguish different inflation models .
the wmap five - year data and the combination of various measurements favor a red power spectrum , which could put tight constraints on the kklmmt model . plane .
right : @xmath0 vs. @xmath42 for different number of e - folds @xmath21.,width=316 ] now we plot the correlation of @xmath128 in the left panel of fig .
[ model3 ] to see what is the trajectory of the model when varying the parameter @xmath0 . in this figure , by decreasing the parameter @xmath0 , we show that the trajectory of the model goes from the upper - right corner to the lower - left corner in the @xmath128 plane .
so , only @xmath129 could give both the red - tilted scalar spectrum and the small field manner .
this is due to the fact that a larger parameter @xmath0 will make the field rolling too fast , which might make the slow - roll parameter @xmath44 be positive .
therefore , the red - tilted power spectrum requires @xmath129 .
we also plot @xmath0 vs. @xmath130 for different number of e - folds @xmath21 in the right panel of fig .
[ model3 ] . in this figure
, it is easy to see that different number of @xmath20-folds @xmath21 can make little influence on the curves , and the red - tilted spectrum roughly requires @xmath131 . summarizing the above discussions
, we see that the spectral index can put the tightest constraints on the kklmmt model .
this is obvious because the spectral index is the very quantity that has been precisely measured while other quantities only get loose upper limits from the observations .
the aim of making these discussions is to see from various angles of view what constraints these observational quantities can bring to the model . from the above analysis , we get to know that the current observational bound on the parameter @xmath0 is roughly @xmath132 further constraints from the wmap five - year results will be discussed in the next subsection .
we now pause to discuss the physical effect of the parameter @xmath0 .
if @xmath0 goes larger , the potential @xmath133 will become steeper , which leads the inflaton to roll faster than the small @xmath0 case .
thus , the change magnitude of the inflaton field during inflation will be greater . at the same time
, the @xmath36 and @xmath98-branes will have greater momentum to collide , producing more @xmath134-strings and @xmath135-strings .
therefore , constraining the parameter @xmath0 is actually constraining the shape of the potential and the dynamics of inflation .
the wmap three - year data give a rather red spectrum , i.e. @xmath136 ( 1@xmath41 cl ) for wmap3 data only , and @xmath137 ( 1@xmath41 cl ) for the combination of wmap3 and sdss data .
therefore , such a red spectrum requires the parameter @xmath0 in the kklmmt to be fine - tuned . in @xcite , it has been shown that , with the wmap three - year data , the constraints on the parameter @xmath0 are @xmath138 at the level of 1@xmath41 and @xmath139 at the level of 2@xmath41 for wmap3 only ; @xmath140 at the level of 2@xmath41 for wmap3 + sdss .
so , the fine - tuning for the parameter @xmath0 is needed , according to the constraints from the wmap three - year results ( see also @xcite ) .
however , the scalar spectral index derived from the wmap five - year data is relatively blue comparing to that of wmap3 , though it is still red - tilted .
( [ nsnumber1 ] ) and ( [ nsnumber2 ] ) for the results of wmap5 .
therefore , we expect that these results could relax the fine - tuning of @xmath0 to some extent . is forbidden since in this case the @xmath0 term in the potential tends to push the d3-brane out of the throat such that the inflation will not happen .
left : wmap five - year data only ; right : wmap five - year data combined with snia and bao data .
, width=316 ] we plot the spectral index @xmath130 as a function of the number of @xmath20-folds @xmath21 in fig .
[ ns2 ] . for comparison
, we also show the observational results of the spectral index , from the wmap five - year data . in this figure , it is easy to see that the larger the parameter @xmath0 , the greater the spectral index .
we also find that the curves are very sensitive to the parameter @xmath0 , i.e. , the curves of the spectral index will change significantly even though the parameter @xmath0 only changes a little , e.g. , from @xmath141 to @xmath89 .
note that @xmath142 is forbidden since in this case the @xmath0 term in the potential tends to push the d3-brane out of the throat such that the inflation will not happen .
for the boundary curve corresponding to @xmath143 , the central value of @xmath42 gives a reasonable value of the @xmath20-folding number @xmath21 ; however , when changing @xmath0 to @xmath89 , the model could not provide the central value of @xmath130 within its reasonable range of the number of @xmath20-folds , which implies that the parameter @xmath0 tends to be fine - tuned confronting a red - tilted spectrum of the primordial perturbation .
@xmath144{fit1.eps } & \includegraphics[scale=0.85]{fit2.eps } \\ \end{array}$ ] confronting the wmap five - year results of the spectral index , eqs .
( [ nsnumber1 ] ) and ( [ nsnumber2 ] ) , the correlation between the parameter @xmath0 and the @xmath20-folding number @xmath21 can also be derived , see fig .
[ fig : nbeta ] . from this figure
, one can clearly read off the allowed region in the parameter - space by the wmap five - year data .
note that the area of @xmath145 and @xmath38 is a rational region for the kklmmt model .
therefore , according to the wmap five - year results of the spectral index , we see that the kklmmt model can fit well with both the data of wmap only and wmap + bao + sn within 1@xmath41 range .
for wmap only , we derive @xmath146 at the level of 1@xmath41 and @xmath147 at the level of 2@xmath41 . for wmap + bao + sn , we derive @xmath148 at the level of 1@xmath41 and @xmath149 at the level of 2@xmath41 .
so , comparing to the wmap five - year results , we find that the value of the parameter @xmath0 is relaxed to @xmath150 at the level of 2@xmath41 . the problem of fine - tuning of @xmath0 is alleviated to a certain extent when confronting the wmap five - year data .
this is , without doubt , a good news for the kklmmt model . for the running of the spectral index , according to the wmap five - year data
, one finds an upward shift from the three - year result , @xmath151 ( wmap5 + bao + sn ) .
, may indicate that there is no evidence for the running of the spectral index , as pointed out in @xcite . ] the kklmmt model , generally , can not yield a seeable running of the spectral index .
however , for @xmath152 , a large running is also possible in the kklmmt model .
unfortunately , even though taking a big @xmath0 , the kklmmt model can not provide a running compatible with the current wmap result for the running , see fig .
[ fig : running ] . of course
, the future high precision observations might make a more precise measure for the running of the spectral index , and might demonstrate explicitly that there is no running of the spectral index .
furthermore , we also compare the angular power spectrum of the kklmmt model with the wmap five - year data , by employing the camb code @xcite , see fig .
[ angular ] . in the right panel of fig .
[ angular ] , we see that the smaller the value of @xmath0 is , the larger the first peak amplitude will be , and , obviously , @xmath153 is not able to fit the power spectrum at all . thus , the current data of the cmb angular power spectrum also indicate that the parameter @xmath0 should be taken as @xmath154 , which is consistent with the constraint from the spectral index .
finally , let s discuss the prospects for the future measurements of the tensor - to - scalar ratio , @xmath8 , in light of the existing constraint on @xmath42 . from the above analysis , we know that the current constraints on @xmath42 put a bound on the kklmmt model of order @xmath155 or so .
such a small value of @xmath0 leads to a negligible @xmath8 , less than @xmath156 , see fig .
[ tensor1 ] .
this means that if current bounds on @xmath42 do not shift appreciably , any observation of primordial tensor modes will rule out the kklmmt model entirely .
therefore , the future measurements of @xmath8 might provide a disproof for the kklmmt model .
in this paper , we have studied brane inflation with the wmap five - year data .
we first considered a general brane inflation model in which the problem of dynamic stabilization is neglected .
furthermore , we examined the kklmmt model by using the wmap five - year results . for the general inflation model ,
we show that , according to the wmap five - year data , the model survives at the level of 1@xmath41 .
this conclusion is different from that of a previous work @xcite which is based on the analysis of the wmap three - year data . in @xcite , with the wmap three - year data , it was indicated that the brane inflation model for @xmath43 ( d3-brane case ) is near the boundary of the wmap3 only and can not fit the wmap3+sdss data at the level of 1@xmath41 ; the model with @xmath34 ( d5-brane case ) is outside the range allowed by wmap3 only or wmap3+sdss at the level of 1@xmath41 .
therefore , we find that the wmap five - year data save the brane inflation model to the level of @xmath122 . for the kklmmt model ,
we first discuss how the various observables can bring constraints on the parameter @xmath0 of the model .
then , we compare the model to the wmap five - year results . in @xcite , by using the wmap three - year data , the authors find that the kklmmt model can not fit wmap3+sdss data at the level of 1@xmath41 and a fine - tuning , at least eight parts in a thousand , is needed at the level of 2@xmath41 .
however , in this work , we show that the kklmmt model is consistent with both the data of wmap5 only and of wmap5+bao+sn at the level of @xmath122 .
moreover , comparing to the wmap five - year results , we find that the value of the parameter @xmath0 is relaxed to @xmath150 at the level of 2@xmath41 .
therefore , the problem of fine - tuning of @xmath0 is alleviated to a certain extent when confronting the wmap five - year data .
this is definitely a good news for the kklmmt model .
this paper only discusses the the simplest realistic brane inflationary model , namely the kklmmt model , in which the usual slow - roll scenario is employed , but does not involve the dirac - born - infeld ( dbi ) scenario , in which the rolling of the inflaton is albeit slow but relativistic . in the dbi model , large tensor mode and/or non - gaussianity may emerge , providing a possible stringy signature .
the dbi model has been investigated in detail in @xcite by using the wmap five - year data .
one of us ( yzm ) would like to thank anne - christine davis , george efstathiou , hiranya peiris , qing - guo huang , simeon bird and yi wang for helpful discussions , also thank duncan hanson and damien quinn for computer help .
this work was supported by grants from the cambridge overseas trust and the studentship from trinity college , cambridge , and the natural science foundation of china .
99 a. h. guth , phys .
d. 23 ( 1981 ) 347 ; a. d. linde , phys .
b. 108 ( 1982 ) 389 ; a. albrecht and p. j. steinhardt , phys .
48 ( 1982 ) 1220 ; for earlier attempts on an inflationary model , see , a. a. starobinsky , phys .
b * 91 * , 99 ( 1980 ) .
g. f. smoot _
et al . _ ,
j. 396 ( 1992 ) l1 ; c. l. bennett _
et al . _ ,
j. 464 ( 1996 ) l1 , astro - ph/9601067 .
c. l. bennett _
_ , astrophys
. j. suppl . 148
( 2003 ) 1 , astro - ph/0302207 ; d. n. spergel _ et al .
_ , astrophys
. j. suppl . 148
( 2003 ) 175 , astro - ph/0302209 ; h. v. peiris _ et al .
_ , astrophys
. j. suppl . 148
( 2003 ) 213 , astro - ph/0302225 .
s. kachru , r. kallosh , a. linde and s. p. trivedi , phys .
d * 68 * , 046005 ( 2003 ) [ hep - th/0301240 ] .
s. kachru , r. kallosh , a. linde , j. m. maldacena , l. mcallister and s. p. trivedi , jcap 0310 , ( 2003 ) 013 , hep - th/0308055 .
r. bean , s. e. shandera , s. h. henry tye and j. xu , jcap * 0705 * , 004 ( 2007 ) [ arxiv : hep - th/0702107 ]
. l. lorenz , j. martin and c. ringeval , jcap * 0804 * , 001 ( 2008 ) [ arxiv:0709.3758 [ hep - th ] ] .
l. alabidi and j. e. lidsey , phys .
d * 78 * , 103519 ( 2008 ) [ arxiv:0807.2181 [ astro - ph ] ] .
b. s. mason , _ et .
al _ apj , 591 ( 2003 ) 540 ; j. l. sievers , _ et .
al _ , apj , 591 ( 2003 ) 599 ; j. l. sievers , _ et .
al _ apj , 660 ( 2007 ) 976 ; t. j. pearson , _ et .
al _ , apj , 591 ( 2003 ) 556 ; a. c. s. readhead , _ et .
al _ apj , 609 ( 2004 ) 498 ; c. dickinson , _ et
. al _ mnras , 353 ( 2004 ) 732 ; c. l. kuo , _ et .
al _ apj , 600 ( 2004 ) 32 ; c. l. kuo , _ et .
al _ apj , 664 ( 2007 ) 687 ; j. e. ruhl , _ et .
al _ apj , 599 ( 2003 ) 786 ; t. e. montroy , _ et .
al _ apj , 647 ( 2006 ) 813 ; f. piacentini , _ et .
al _ apj , 647 ( 2006 ) 833 .
u. l. pen , u. seljak and n. turok , phys .
79 , ( 1997 ) 1611 , astro - ph/9704165 ; b. allen , r. r. caldwell , s. dodelson , l. knox , e. p. s. shellard and a. stebbins , phys .
79 , ( 1997 ) 2624 , astro - ph/9704160 ; a. albrecht , r. a. battye and j. robinson , phys . rev .
d 59 ( 1999 ) 023508 , astro - ph/9711121 .
s. e. shandera and s. h. tye , jcap * 0605 * , 007 ( 2006 ) [ arxiv : hep - th/0601099 ] .
m. wyman , l. pogosian and i. wasserman , phys .
d. 72 , ( 2005 ) 023513 ; m. wyman , l. pogosian and i. wasserman , phys .
d. 73 , ( 2006 ) 089905 , erratum , astro - ph/0503364 ; a. a. fraisse , jcap 03 ( 2007 ) 008 , astro - ph/0603589 ; l. pogosian , i. wasserman and m. wyman , astro - ph/0604141 . | in this paper , we revisit brane inflation models with the wmap five - year results .
the wmap five - year data favor a red - tilted power spectrum of primordial fluctuations at the level of two standard deviations , which is the same as the wmap three - year result qualitatively , but quantitatively the spectral index is slightly greater than the three - year value .
this result can bring impacts on brane inflation models . according to the wmap five - year data , we find that the kklmmt model can survive at the level of one standard deviation , and the fine - tuning of the parameter @xmath0
can be alleviated to a certain extent at the level of two standard deviations . |
the concepts of baryonic matter in a cold dilute phase with clustering of nucleons in the lightest nuclei @xmath11he , @xmath6he , @xmath12h and @xmath6h have been developed in the last decade @xcite .
theoretical developments carried out in this direction orient towards the study of cluster groups as integral quantum systems and give motivation to a new generation of experiments on cluster spectroscopy @xcite . since the macroscopic cluster states can play the role of an intermediate phase in astrophysical processes ,
these studies assume the significance going beyond the framework of the nuclear structure problems @xcite .
the method of nuclear track emulsion provides a uniquely complete observation of multiple fragment systems produced in dissociation of relativistic nuclei .
approximate conservation of the initial momentum per nucleon by relativistic fragments is used in the kinematical analysis of the events to compensate the lack of momentum measurements .
the fragmenting system excitation can be defined as q = m@xmath13-m , where m@xmath13 is the invariant mass and the projectile mass or total fragment mass .
the value m@xmath13 is defined by the relation m@xmath14 = ( @xmath15p@xmath16)@xmath12=@xmath15(p@xmath17p@xmath18 ) , where p@xmath19 4-momenta fragments i and the k , determined in the approximation of the conservation of the primary momentum value per nucleon .
the most valuable events of coherent dissociation of nuclei in narrow jets of light and the lightest nuclei with a net charge as in the initial nucleus , occurring without the formation of fragments of the target nuclei and mesons ( the so - called whitestars ) @xcite , comprise a few percent among the observed interactions .
the data on this phenomenon are fragmented , and the interpretation is not offered . the dissociation degree of light o , ne , mg and si , and as well as heavy au , pb and u nuclei may reach a complete destruction to light and the lightest nuclei and nucleons , resulting in cluster systems of an unprecedented complexity .
the dissociation dynamics of heavy nuclei can be grounded on dissociation peculiarities established for light nuclei .
an extensive collection of photographs of such interactions is gathered by the becquerel collaboration @xcite . despite the fact that the potential of the relativistic approach to the study of nuclear clustering is recognized long ago , e - experiments were not be able to get closer to the required detailed observation of the relativistic fragment ensembles .
the related pause has led to the proposal to irradiate nuclear track emulsion in the jinr nuclotron beams of the whole family of 1.2 a gev light nuclei , including radioactive ones @xcite .
studies with relativistic neutron - deficient nuclei have special advantages due to more complete observations .
the dissociation features of @xmath0be , @xmath1c , and @xmath2n nuclei , which are the sources of basic cluster configurations , will be presented in the present paper .
in order to justify the application of the relativistic fragmentation in the study of n@xmath7-systems @xcite it was suggested to investigate the dynamics of the formation of @xmath7-particle pairs at a high statistical level and under the simplest conditions ( without combinatorial background ) which are provided in the relativistic fragmentation @xmath0be@xmath52@xmath7 @xcite .
the secondary @xmath0be beam was obtained by fragmentation of accelerated @xmath8b nuclei @xcite . when scanning the exposed emulsion 500 events @xmath0be@xmath52@xmath7 in a fragmentation cone of 0.1 rad have been found .
about 81% @xmath7-pairs form roughly equal groups on @xmath20 : narrow(0 @xmath21 10.5 mrad ) and wide(15.0 @xmath22 45.0 mrad ) ones .
the @xmath23 pairs are consistent with @xmath4be decays from the ground state 0@xmath3 , and pairs @xmath24 - from the first excited state 2@xmath3 .
the @xmath23 and @xmath24 fractions are equal to [email protected] and [email protected] .
these values are well corresponding to the weights of the @xmath4be 0@xmath3 and 2@xmath3 states @xmath26 = 0.54 and @xmath27 = 0.47 in the two - body model n - @xmath4be , used to calculate the magnetic moment of the @xmath0be nucleus @xcite .
for the coherent dissociation @xmath0be@xmath52@xmath7+n , the average value of the total @xmath7-pair transverse momentum is equal to @xmath28 80 mev / c , which can be assigned to the average transverse momentum carried away by neutrons . for the @xmath0be coherent dissociation through the @xmath4be 0@xmath3 and 2@xmath3 states
there is no differences in the values @xmath29 , which points to a cold fragmentationmechanism .
the whole complex of these observations may serve as an evidence of the simultaneous presence of the @xmath4be 0@xmath3 and 2@xmath3 states with similar weights in the ground state of the nucleus @xmath0be .
one can expect that the pattern established for the @xmath30be @xcite and @xmath4b @xcite nuclei is reproduced for nucleus @xmath0c with the addition of one or two protons .
in addition , the emergence of a 3@xmath6he ensemble becomes possible .
an intriguing hypothesis is that in the nuclear astrophysical processes the 3@xmath6he system can be a 3@xmath7-process analog .
a secondary beam , optimized for @xmath0c nucleus selection was formed by fragmentation of accelerated @xmath2c nuclei @xcite .
it was important in this irradiation to avoid overexposure by the accompanying flux of @xmath6he nuclei .
the intensity ratio of the nuclei with charges z@xmath31 = 6 and 2 amounted to 1 : 10 .
this factor has limited statistics and made the scan for @xmath0c interactions much more labor demanding . among the total number of whitestars n@xmath32 , detected in this exposure , 15 events @xmath0c@xmath10b+p and 16 events
@xmath30be+2p are found .
statistics in the channels 2he+2h ( 24 ) , he+4h ( 28 ) and 6h ( 6 ) well corresponds to the @xmath30be core dissociation .
the event fraction @xmath0c@xmath53@xmath6he ( 16 ) was found to be the same as that of the channels @xmath0c@xmath10b+p and @xmath30be+2p .
the latter fact can point to a significant admixture of a virtual 3@xmath6he state in the @xmath0c ground state .
this component can give a contribution to the @xmath0c magnetic moment , which has an abnormal value in terms of the shell model @xcite .
the @xmath8c nucleus is the only example of the system , which has the super - boromean properties , since the removal of one of the four clusters in the 2@xmath7 + 2p structure ( threshold 3.8 mev ) leads to an unbound state .
the particular feature of the @xmath2n nucleus consists in the low proton separation threshold ( 600 kev ) .
furthermore , the dissociation can occur through the channels @xmath7+@xmath4b ( 8 mev ) , p+@xmath30be+@xmath7 , as well as into more complicated ensembles with the @xmath30be core break .
generation of @xmath2n and @xmath8c nuclei is possible in charge exchange and fragmentation reactions of accelerated @xmath2c nuclei @xcite .
the charge to weight ratio z@xmath31/a@xmath31 differs by only 3% for these nuclei , while the momentum acceptance of the separating channel is 2 - 3% .
therefore , their separation is not possible , and the @xmath2n and @xmath8c nuclei are simultaneously present in the secondary beam , forming a so - called beam cocktail .
the contribution of @xmath2n nuclei is small in respect to @xmath8c ones in accordance with the cross sections for charge transfer and fragmentation reactions .
also , the beam contains @xmath30be nuclei , differing by z@xmath31/a@xmath31 from @xmath2n nuclei only by 2% . due to
the momentum spread @xmath6he nuclei can penetrate in the separating channel . for neighboring @xmath4b , @xmath0c and @xmath33c nuclei
the difference by z@xmath31/a@xmath31 from @xmath2n is about 10% , which leads to suppression in these isotopes .
identification of @xmath2n nuclei can be performed by @xmath34-electron counting along the beam tracks . in the @xmath8c case ,
relying on the charge topology of the produced whitestars it is necessary to be sure that the neighboring carbon isotope contribution is small .
these considerations provided the justification to expose nuclear track emulsion in a mixed beam of @xmath2n , @xmath8c and @xmath30be nuclei . the initial scanning phase consisted in visual search of beam tracks with charges z@xmath31=1 , 2 and z@xmath352 .
the ratio of beam tracks with charges z@xmath31= 1 , 2 and z@xmath362 is found to be equal @xmath37 1 : 3 : 18 .
thus , the contribution of @xmath6he nuclei dramatically decreased compared with the @xmath0c irradiation , which radically raised the event search efficiency .
the presence of fragments z@xmath382 makes the charge identification of beam z@xmath31 and secondary z@xmath39 tracks necessary . for the calibration the average density of @xmath34-electrons n@xmath40 was measured along the beam tracks , which produced the whitestars 2he+2h , 2he and he+2h , and also stars with fragments z@xmath382 ( @xmath2n candidates ) .
thus , the correlation between the charge topology @xmath15z@xmath39 and n@xmath40 was established which permitted to determine beam track charges z@xmath31 and the fragment charges z@xmath382 . for whitestars
n@xmath32 with charge topology @xmath15z@xmath39=6 the most probable channel is represented by 91 events 2he+2h , which might be expected for the isotope @xmath8c .
the channel he+4h is found to be suppressed ( 14 events ) , as in the @xmath8c case it is required to overcome the high threshold of the @xmath7-cluster break up . in this irradiation
20 whitestars with z@xmath31=7 and @xmath15z@xmath39=7 are found , corresponding to the dissociation of @xmath2n nuclei .
there are the following channels among them : c+h ( 1 ) , @xmath30be+he+h ( 2 ) , @xmath30be+3h ( 4 ) , @xmath4b+2h ( 3 ) , 3he+h ( 2 ) , 2he+3h ( 6 ) , he+5h ( 3 ) .
thus , half of the events contain a fragment z@xmath38 2 , clearly differing from the cases of nuclei @xmath41n @xcite and @xmath8c .
the unbound @xmath4be nucleus plays the role of the core in the @xmath8c structure , which should be manifested in the fragmentation intensity @xmath8c@xmath10be .
distribution @xmath7-pairs in the 91 whitestars 2he+2h on the excitation energy q@xmath42 is presented in fig .
1 . in 30 events
the q@xmath42 value does not exceed 500 kev ( inset in fig .
1 ) . for them , the average value is @xmath43110@xmath2520 kev and the mean - square scattering @xmath44=40 kev , which well corresponds to the decays of the @xmath4be 0@xmath3 ground state .
the @xmath4be 0@xmath3 contribution is approximately the same as for the neighboring cluster nuclei
. the unbound @xmath0b nucleus can be another major product of the @xmath8c coherent dissociation .
2 shows the distribution of whitestars 2he+2h on the excitation energy q@xmath45 , defined by the difference of the invariant mass of the three fragments 2@xmath7+p and the mass of the proton and the doubled @xmath7-particle mass .
the q@xmath45 values for one of two possible 2@xmath7+p triples do not exceed 500 kev in 27 events ( inset in fig .
the average value for these triples is @xmath46=250@xmath2515 kev with rms @xmath44=74 kev .
these values well correspond to the @xmath0b ground state decay via the channel p+@xmath4be ( 0@xmath3 ) having the published values of energy 185 kev and width ( [email protected] ) kev@xcite . in the region limited by
q@xmath471 mev ane time both triples correspond to the decay of the nucleus @xmath0b . in all other @xmath0b cases
one of q@xmath45 is above 500 kev .
in addition , excitations @xmath7
+ 2p are studied on the remaining statistics of whitestars 2he+2h beyond @xmath0b decays . in the spectrum of q@xmath45 , there is no clear signal of @xmath48be decays @xcite , and its estimated contribution does not exceed 20% .
this aspect deserves further analysis taking the proton angular correlations into account .
the work was supported by grants of the russian foundation for basic research ( 96 - 1596423 , 02 - 02 - 164 - 12 , 03 - 02 - 16134 , 03 - 02 - 17079 , 04 - 02 - 17151 . 04 - 02 - 16593 and 09 - 02 - 9126
ct - a ) and grants of the jinr plenipotentiaries of the republic of bulgaria , the slovak republic , the czech republic and romania in the years 2002 - 2010 . | the dissociation features in nuclear track emulsion of @xmath0be , @xmath1c , and @xmath2n nuclei of 1.2 a gev energy are presented .
the data presented for the nucleus @xmath0be can be considered as evidence that there is a core in its structure in the form of 0@xmath3 and 2@xmath3 states of the @xmath4be nucleus having roughly equal weights .
events of coherent dissociation @xmath0c@xmath53@xmath6he associated with the rearrangement of the nucleons outside the @xmath7-clustering are identified .
a pattern of the charge fragment topology in the dissociation of @xmath8c and @xmath2n nuclei is obtained for the first time .
contribution of the unbound nucleus decays to the cascade process @xmath8c@xmath9b@xmath10be is identified . |
i have been asked to describe quantum fluctuations in cosmology , which i find a fascinating topic .
it is a dramatic demonstration that the quantum theory that was developed by studying the hydrogen atom can be applied on larger and larger scales .
here we are applying quantum theory to the universe in its entirety , at time scales of order @xmath0 second , and it all sounds incredibly fantastic .
but the shocking thing is that it works , at least in the sense that it gives answers for important questions that agree to very good precision with what is actually measured .
in addition to discussing the density perturbations that we can detect , however , i want to also discuss another important aspect of quantum fluctuations : specifically , quantum fluctuations in cosmology appear , in almost all our models , to lead to eternal inflation and an infinite multiverse .
this is a rather mind - boggling concept , but given our success in calculating the fluctuations observed in the cosmic microwave background ( cmb ) , it should make good sense to consider the other consequences of quantum fluctuations in the early universe .
thus , i think it is time to take the multiverse idea seriously , as a real possibility .
the inhomogeneities that lead to eternal inflation are nothing more than the long - wavelength tail of the density perturbations that we see directly in the cmb .
the idea that quantum fluctuations might be the origin of structure in the universe goes back at least as far as a 1965 paper by sakharov @xcite . in the context of inflationary models , the detailed predictions are model - dependent , but a wide range of simple models give generic predictions which are in excellent agreement with observations . in this section
i will give a pedagogical explanation of how these predictions arise , based on the time - delay formalism that was used in the paper i wrote with s .- y .
pi @xcite .
this formalism , which we learned from stephen hawking , is the simplest to understand , and it is completely adequate for the dominant perturbations in single - field , slow - roll inflation .
, has not ( to my knowledge ) been confirmed in any modern paper .
however , the fact that @xmath1 is proportional to @xmath2 has been confirmed , showing that the 1981 paper by mukhanov and chibisov did correctly calculate what we now call @xmath3 ( as was pointed out in ref . ) . ]
more sophisticated approaches are needed , however , to study multifield models or models that violate the slow - roll approximation , or to study extremely subdominant effects in single - field , slow - roll models . even for multifield inflation , however , some of the simplicity of the time - delay formalism can be maintained by the use of the so - called @xmath4 formalism @xcite .
there are a number of reviews @xcite and textbooks @xcite that give a much more thorough discussion of density perturbations in inflationary models than is appropriate here .
inflation @xcite takes place when a scalar field has a large potential energy density . a straightforward application of noether s theorem @xcite gives the energy - momentum tensor of a canonically normalized scalar field as @xmath5 where @xmath6 where @xmath7 .
so , as long as the energy of the state is dominated by @xmath8 , guarantees that the pressure is large and negative .
einstein s equations imply that negative pressure creates repulsive gravity , so any state whose energy is dominated by the potential energy of a scalar field will drive inflation .
there are two basic scenarios one where @xmath9 starts at the top of a hill ( new inflation @xcite ) , and one where it starts high on a hill and rolls down ( chaotic inflation @xcite ) ; see .
either scenario is successful , and for the density perturbation calculation we can treat them at the same time .
we use comoving coordinates , with a background metric describing a flat friedmann - robertson - walker ( frw ) universe : @xmath10 where @xmath11 is the scale factor .
objects moving with the expansion of the universe are at rest in this coordinate system , with the expansion described solely by @xmath11 : when @xmath11 doubles , all the distances in the universe double . in this metric the klein - gordon equation for a scalar field is given by @xmath12 where an overdot indicates differentiation with respect to time @xmath13 , and @xmath14 is the laplacian , @xmath15 , with respect to the coordinates @xmath16 .
the equation is identical to the klein - gordon equation in minkowski space , except that there is a drag term , @xmath17 , which can be expected , since the energy density must fall if the universe is expanding .
in addition , each spatial gradient is modified by @xmath18 , which converts the derivative to the current scale of spatial distance .
the hubble expansion rate @xmath19 is given by the friedmann equation for a flat universe , @xmath20 in this language , the repulsive effect of the negative pressure that was mentioned above can be seen in the equation for the acceleration of the expansion , @xmath21 if @xmath22 , as one finds when @xmath8 dominates , this equation gives @xmath23 .
using an assumption called the slow - roll approximation , which is valid for a large range of inflationary models , we can ignore the @xmath24 term of and the @xmath25 term in .
in addition , at sufficiently late times the laplacian term can be neglected , since it is suppressed by @xmath26 .
we are then left with a very simple differential equation , @xmath27 which has a one - parameter class of solutions . that one parameter is itself trivial
it is a time offset .
given one solution @xmath28 , the general solution can be written as @xmath29 , where @xmath30 is independent of @xmath13 . since the differential equation ( [ ahg : eq9 ] ) has no spatial derivatives , @xmath30 can depend on position , so the most general solution can be written as @xmath31 since we are interested in developing a first order perturbation theory , we can expand about @xmath28 , @xmath32 so @xmath33 even though the numerator and denominator of the above expression both depend on time , the quotient does not . thus at late times ( within the inflationary era )
times late enough for to be accurate the nonuniformities of the rolling scalar field are completely characterized by a time - independent time delay.the description of the perturbations at late times by a time - independent time delay @xmath34 is in fact much more robust than the approximation that @xmath24 can be neglected .
it is a consequence of the hubble drag term , and will hold at sufficiently late times in any single - field model for which the slow - roll approximation is valid for more than a few @xmath35-folds . to see this , consider , with @xmath36 taken to be an arbitrary function of @xmath9 and @xmath37 .
we will neglect the laplacian term , since it is suppressed by @xmath26 , and we are interested in late times .
then , for each value of @xmath38 there is a two - parameter class of solutions to this second order ordinary differential equation . to see the effect of the damping , suppose that we know the unperturbed solution , @xmath28 , and a nearby solution , @xmath39 , where @xmath40 is to be treated to first order
@xmath41 can depend on @xmath38 , but we suppress the argument because we consider one value of @xmath38 at a time .
we then find that @xmath40 and @xmath42 obey the same differential equation .
if we construct the wronskian @xmath43 , we find that @xmath44 the solution to which is @xmath45 thus @xmath46 falls off roughly as @xmath47 or faster ( @xmath48 ) , and so can be neglected after just a few @xmath35-folds of expansion . then note that @xmath49 while in the slow - roll regime @xmath50 is approximately constant from one can show that the fractional change in @xmath50 during one hubble time ( @xmath51 ) is approximately @xmath52 , as defined in eqs .
( [ ahg : eq14 ] ) and ( [ ahg : eq15 ] ) . thus the time derivative of the ratio @xmath53 falls off as @xmath54 or faster , implying that the time delay rapidly approaches a fixed value .
the time - delay description remains accurate throughout the reheating process , even though the slow - roll conditions will generally fail badly at the end of inflation , when the scalar field starts to oscillate about the bottom of the potential well .
the time delay is maintained because @xmath41 and @xmath55 continue to obey the same linear differential equation .
thus if @xmath56 at the end of the slow roll period , then @xmath41 will remain proportional to @xmath55 for all later times . this argument , which generalizes an argument in ref .
, is in contradiction with ref .
, where it is argued that the time delay persists to the end of inflation only under very stringent assumptions about the potential . the argument of ref . , however , is really a discussion of the validity of , but we have seen that the time delay is preserved even when fails . ]
it is useful to define a dimensionless measure of the time delay , @xmath57 which can be interpreted as the number of @xmath35-folds of inflation by which the field is advanced or retarded . to justify the slow - roll approximation , we must adopt restrictions on the form of the potential energy function @xmath8 .
the slow - roll approximation is equivalent to saying that the field @xmath9 evolves approximately at the drag - force limited velocity , where the drag force equals the applied force , with inertia playing only a negligible role .
( this would be called the terminal velocity , except that it can change slowly with time . ) from the first two terms of one can see that the velocity approaches the drag - limited value with a time constant of order @xmath51 .
thus , for the field to evolve at the drag - limited velocity , it is essential that neither the drag coefficient nor the applied force changes significantly during a time of order @xmath51 .
thus we want to insist that @xmath58 , and that @xmath59 . using to approximate @xmath60 ,
these two conditions can be expressed in terms of the two slow - roll parameters @xcite @xmath61 where a prime denotes a derivative with respect to @xmath9 .
note that these slow - roll conditions do not by themselves guarantee that @xmath9 will evolve at drag - limited velocity , because a large initial velocity will take time before it approaches the drag - limited value .
but the slow - roll conditions do guarantee that for times long compared to @xmath51 , @xmath9 will evolve at very nearly the drag - limited velocity .
[ ahg : fig2 ] to proceed , i will make two approximations that will simplify the problem enormously , but which are nonetheless extremely accurate for single - field slow - roll inflation .
first , we will neglect all perturbations of the metric until the time when inflation ends .
that is , until inflation ends we treat the scalar field as a quantum field in a fixed de sitter space background .
thus , we will be ignoring the fluctuations in the energy - momentum tensor of the scalar field , since we are not allowing them to perturb the metric .
however , we will calculate the fluctuations in the scalar field itself , as described by the time delay @xmath62 . since the scalar field is driving the inflation , the time delay @xmath34 measures the variation in the time at which inflation ends at different places in space .
the amount of energy that is released at the end of inflation is much larger than the energy - momentum tensor fluctuations during inflation , so the spatial variation of the timing of this energy release becomes the dominant source of the density perturbations that persist at later times . to describe this release of energy ,
we make our second approximation .
we will treat the ending of inflation as instantaneous .
we will assume that the potential energy of the inflaton field is converted instantaneously into the thermal radiation of effectively massless particles , beginning the radiation - dominated era of cosmological history .
i will give heuristic justifications for these approximations , but i am not aware of a more rigorous justification that can be explained without developing an understanding of what happens when these approximations are avoided , and then the problem requires a much more detailed analysis .
such analyses have of course been done and are even described in textbooks .
the answer that we will obtain agrees with the textbooks @xcite , for the single - field slow - roll case , down to the last factor of @xmath63 . while the textbooks corroborate the answer that we will obtain , the methods are sufficiently different so that very little light is shed on the approximations described here . in a future publication@xcite ,
i will attempt a more detailed justification . for a given theory we can calculate @xmath28 by solving an ordinary differential equation ,
so reduces the problem of calculating @xmath34 to that of calculating the fluctuations of the scalar field , @xmath64 .
this is a problem in quantum field theory , albeit quantum field theory in curved spacetime .
the calculations closely resemble the familiar quantum field theory calculations for minkowski space , but there is one point in the calculation where cosmology rears its head .
while minkowski space has a well - defined vacuum state , which is the starting point for most calculations , it is less clear what quantum state should be used to describe the fields evolving in de sitter space , where the time dependence prevents the existence of a conserved total energy . in principle the quantum state is determined by the initial conditions for the universe , about which we know very little . however , while we do not know the quantum state of the early universe , there is a very natural choice , corresponding at least locally to the concept of a vacuum state . to understand this choice , recall that we are interested in an exponentially expanding space , the de sitter spacetime of inflation , so to a good approximation @xmath65 , where @xmath36 is constant .
if we now treat the inflaton field @xmath66 as a quantum operator , we can as usual consider its fourier transform : @xmath67 for a free quantum field theory in minkowski spacetime , @xmath68 would be the sum of an annihilation operator term for particles of momentum @xmath69 and a creation operator term for particles of momentum @xmath70 , each corresponding to a de broglie wavelength @xmath71 . for an frw spacetime ,
since the fourier transform is defined in terms of the _ comoving _ coordinates @xmath38 , the physical wavelength for a mode @xmath69 is not constant , but is given by @xmath72 in other words , each mode is stretched as the universe expands .
thus , if we follow any mode backwards in time , it will have a shorter and shorter wavelength and a higher and higher frequency .
the hubble expansion rate @xmath36 is approximately constant during inflation , so at very early times @xmath36 is very small compared to the frequency , and hence is negligible .
thus , any given mode behaves at asymptotically early times exactly like a mode in minkowski space , so the `` natural '' initial state is to simply start each mode in its minkowski vacuum state in the asymptotic past .
this is called the bunch
davies vacuum @xcite , and it is identical to what is also called the gibbons hawking vacuum @xcite .
gibbons and hawking developed their description of the vacuum in a completely different formalism , based on the symmetries of the de sitter spacetime ( with @xmath73 ) , but the two vacuum states are identical , as one would hope . the bunch
davies / gibbons hawking vacuum is taken as the starting point for all standard calculations of density perturbations . to discuss the spectrum of fluctuations of a spatially varying quantity such as @xmath64 , which is assumed to be statistically homogeneous and isotropic , cosmologists define a power spectrum @xmath74 by defined in ref .
is related by @xmath75 .
another common normalization , called @xmath76 in ref . and @xmath77 in ref .
, is given by @xmath78 , so @xmath79 . according to ref . ,
@xmath80 and @xmath76 are both called _ the spectrum_. in the context of the curvature perturbation @xmath81 , to be defined below , ref
. defines yet another normalization that remains in common use , @xmath82 . ]
@xmath83 or equivalently @xmath84 when @xmath64 is a quantum field operator , the power spectrum is nothing more than the equal - time propagator , which can be calculated straightforwardly once the vacuum is specified , as described in the previous paragraph . from , @xmath85 can be seen to obey the equation @xmath86 to make use of this equation , we consider its behavior around the time of hubble exit , @xmath87 , when the wavelength is approximately equal to the hubble length , defined more precisely by @xmath88 we assume that the slow - roll conditions of eqs .
( [ ahg : eq14 ] ) and ( [ ahg : eq15 ] ) are valid within several hubble times ( @xmath51 ) of @xmath89 ( but it is okay if they are violated later during the period of inflation , as described in footnote [ ahg : footb ] ) . for @xmath90 ,
the @xmath91 term of becomes insignificant . from
we see that the right - hand - side of has magnitude @xmath92 , where @xmath93 .
thus for times up to and including @xmath87 and a little beyond , we can neglect the right - hand - side and treat @xmath94 as a free , massless , minimally coupled field in de sitter space .
it is then straightforward to show that . ]
@xmath95 \ .
\label{ahg : eq22}\ ] ] the time delay should be calculated at a time slightly beyond @xmath89 , say by a few hubble times , when @xmath96 , when we can neglect the @xmath91 terms in both eqs .
( [ ahg : eq20 ] ) and ( [ ahg : eq22 ] ) . using eqs .
( [ ahg : eq9 ] ) , ( [ ahg : eq12 ] ) , and ( [ ahg : eq14 ] ) , one finds several useful expressions for @xmath97 : @xmath98 since the time delay is evaluated a few hubble times beyond @xmath87 , the quantities @xmath99 , @xmath100 , and @xmath36 appearing in the above expressions should all be evaluated at @xmath28 for @xmath101 .
the distinction between @xmath102 and a few hubble times later is important only at higher order in the slow - roll approximation , since the change in @xmath103 over a hubble time is of order @xmath104 .
note that the fluctuations in @xmath30 are inversely proportional to @xmath104 , which is a consequence of the presence of @xmath105 in the denominator of .
we are now in a position to test the consistency of the first of our approximations ( see ) , the approximation of neglecting the fluctuations in the scalar field energy - momentum tensor .
the major source of these fluctuations is the fluctuation in potential energy caused by the fluctuations in @xmath9 .
thus @xmath106 , so @xmath107 \ .
\label{ahg : eq24}\ ] ] thus , the fractional fluctuations in the mass density are of order @xmath108 times the dimensionless fluctuations @xmath97 of the time delay , so we expect that we can neglect the mass density fluctuations if we are interested in calculating the dominant term in the small @xmath104 limit .
thus , we have reduced the calculation of the density fluctuations to the schematic description of , with the power spectrum of the time delay given by . the role of quantum theory in this calculation
is finished it determined the power spectrum of the time delay .
the rest of the calculation is general relativity and astrophysics . here
i will continue the derivation through the end of inflation , which carries it far enough to compare with the literature and to at least qualitatively understand the observational consequences . to understand the implications of @xmath34 in the cosmic evolution as described by
, it is convenient to switch to a new coordinate system in which the end of inflation happens at @xmath109 everywhere .
just define @xmath110 and @xmath111 .
clearly @xmath112 , so the transformation of the metric of becomes @xmath113 , where @xmath114 with an inverse metric , to first order in @xmath30 , given by @xmath115 in the primed coordinates the phase transition happens sharply at @xmath116 , with a sudden change from @xmath117 in the inflationary phase to @xmath118 in the radiation phase . during the inflationary phase
the energy - momentum tensor is simply @xmath119 , while after the transition it is given by .
the energy - momentum tensor must be covariantly conserved , which means that @xmath120 where @xmath121 denotes a covariant derivative .
the affine connection coefficients @xmath122 will not contain any @xmath123-functions , so @xmath124 can not contain any @xmath123-functions either ; thus @xmath125 must be continuous at @xmath116 .
this implies that @xmath126 the second of these equations can only be satisfied if @xmath127 , because @xmath128 can not vanish , as @xmath129 is timelike , and @xmath130 can not vanish without violating the first equation .
thus , the radiation fluid is necessarily at rest in the frame of reference in which the phase transition occurs simultaneously . requiring @xmath131
, one finds that @xmath132 given the above equation , leads immediately to @xmath133 ; the energy density is conserved across the transition . since the einstein equations are partial differential equations that are second order in time derivatives , the metric and its first time derivative will be continuous across @xmath116 .
so we have now found all the information needed to give a well - defined cauchy problem for the evolution of the model universe , starting at the beginning of the radiation - dominated era . at this point
the perturbations of interest have wavelengths vastly larger than the hubble length , but during the subsequent evolution the hubble length will grow faster than the perturbation wavelength , so later the perturbations will come back inside the hubble length . the description of the perturbations through the time of hubble reentry was given in refs .
, and in many later sources , but for present purposes we will stop here .
at this point we can discuss the validity of the second of our key approximations , the approximation of an instantaneous phase transition .
the actual transition , during which the scalar field rolls down the hill in the potential energy diagram and then oscillates about the minimum and reheats , very likely takes many hubble times to complete .
however , we need to keep in mind that the modes of interest exited the hubble horizon some 50 or 60 hubble times before the end of inflation , which means that at the end of inflation their physical wavelength is of order @xmath134 to @xmath135 , or @xmath136 to @xmath137 , times the hubble length .
thus , even if the phase transition takes @xmath138 hubble times , during this time light would be able to travel less than @xmath139 wavelengths .
thus the phase transition is effectively instantaneous , on the time scale that is relevant for influencing a wave with the wavelengths under consideration .
( of course the reheat energy density that we calculated , @xmath140 , was an artificiality of the instantaneous approximation .
but the calculation can easily be adjusted to account for a lower reheat energy density , which can be found by doing a more accurate , homogeneous calculation . at the end of inflation
we would still obtain , in the primed coordinate system , a radiation fluid that is at rest , with a uniform energy density . )
note that the method used here depended crucially on the assumption that all parts of the universe would undergo the same sequence of events , so that the only difference from one place to another is an overall time offset .
if there were more than one field for which the quantum fluctuations were relevant , then this would not be true , since a fluctuation of one field relative to the other could not be described as an overall time offset .
thus , multifield inflation requires a more sophisticated formalism .
the description we have given so far , specifying the metric and the matter content , is sufficient to calculate the rest of the history , but it is rather complicated .
furthermore , it is equivalent to many other complicated descriptions , related by coordinate transformations .
it is therefore very useful to find a coordinate - invariant way of quantifying the density perturbations .
one convenient approach is motivated by considering the friedmann equation for a universe with spatial curvature , a universe which might be closed , open , or in the borderline case , flat : @xmath141 here @xmath142 is a constant , where positive values describe a closed universe , and negative values describe an open one .
we are interested in describing a perturbation of a homogeneous background universe that is flat , @xmath143 .
thus @xmath144 will on average equal @xmath145 , the value for the background universe , but it will fluctuate about this average .
@xmath36 is normally thought of as part of the global description of the universe , but it has a locally defined analog given by @xmath146 where @xmath147 is the covariant derivative and @xmath129 is the fluid velocity .
( here we will deal only with a radiation fluid , but in a multicomponent fluid @xmath129 can be defined in terms of the total energy - momentum tensor , as described on p. 225 of ref . . )
we can then define @xmath148 \ .
\label{ahg : eq33}\ ] ] this quantity has a property called gauge invariance , which means that its value for any coordinate point @xmath149 is not changed , to first order in the size of the perturbations , by any coordinate transformation that is itself of order of the size of the perturbations . in this case , the gauge invariance follows from the fact that , apart from the factor @xmath150 which is irrelevant for this issue , @xmath151 a quantity that is coordinate - invariant , and which vanishes for the background universe .
( note that coordinate invariance by itself is not enough ; if the quantity varied with time in the background solution , then its value at @xmath149 would change if @xmath13 were redefined by a small amount . )
@xmath152 also has the convenient property that it remains constant as long as spatial derivatives can be neglected , because it is exactly conserved for the case of a homogeneous universe .
we can calculate @xmath151 just after the end of inflation , using the primed coordinate system but dropping the primes .
we have @xmath153 , where @xmath154 and @xmath129 is given by .
this gives @xmath155 , which gives @xmath156 where @xmath157 is the hubble expansion rate during inflation , which we have treated as a constant .
, which could be much smaller .
although it is not obvious , however , the reheat energy density does not affect @xmath152 , which is conserved for wavelengths large compared to the hubble length .
thus , the value we obtained here could have been calculated before reheating began , and is equal to the value that holds long after reheating , whether reheating is fast or slow . to understand the evolution of @xmath152 when @xmath36 changes ,
however , requires a more detailed calculation . because of the factor of @xmath150 in , the value of @xmath152 is in fact sensitive to quantities that are suppressed by factors of @xmath158 . to see the conservation of @xmath152 for long wavelengths , one needs to include the contribution of the auxiliary field @xmath159 defined in footnote [ ahg : footf ] .
this issue will be discussed in more detail in ref . .
] as in , @xmath14 denotes the laplacian operator with respect to the coordinates @xmath16 .
then , given the power spectrum of for @xmath160 , we find a power spectrum for @xmath152 given by @xmath161 in the literature @xmath152 is seldom used , but instead it is much more common to use a variable called the curvature perturbation @xmath162 , for which the usual definition is somewhat complicated .
however , it is shown in ref .
that @xmath163 so @xmath164 where @xmath97 is given by .
this answer agrees precisely with the answers obtained in refs .
, , , and .
in this reference corresponds to @xmath165 , and is given on pp .
482 and 491 . to compare with ref . , note that @xmath166 , given on p. 406 , corresponds to @xmath167 , as described on p. 89 , and that @xmath168 is the reduced planck mass , @xmath169 . to compare with ref . ,
use @xmath170 , where @xmath171 is given on p. 170 . as explained in ref .
, @xmath172 and @xmath81 refer to slightly different quantities , but they agree for wavelengths long compared to the hubble length . to compare with ref . , note that @xmath173 , and that the equation for @xmath174 describes the conditions just after the end of inflation , with the scale factor @xmath175 , @xmath176 .
the quantum fluctuations are quantified in this paper by @xmath177 , with @xmath178 , in agreement with of the current paper . ]
the wmap seven - year paper @xcite quotes @xmath179 , where @xmath180 and @xmath181 .
( the wmap papers use @xmath182 for @xmath183 . )
if these fluctuations come from single - field slow - roll inflation , we can conclude from eqs .
( [ ahg : eq37 ] ) and ( [ ahg : eq23 ] ) that at the time of hubble exit , @xmath184 where @xmath185 gev is the reduced planck mass .
from eqs . ( [ ahg : eq37 ] ) and ( [ ahg : eq23 ] ) we can also deduce how the intensity of the fluctuations varies with @xmath142 , a relation which is parameterized by the scalar spectral index @xmath186 , defined by @xmath187 to evaluate this expression for slow - roll inflation , we use the last expression in for @xmath188 ; we recall that the expression is to be evaluated at @xmath189 ( or some fixed number of hubble times later ) , and find that leads to @xmath190 . then using to write @xmath191 ,
these equations can be combined to give @xcite @xmath192 the wmap seven - year paper @xcite quotes @xmath193 .
this result suggests that the slow roll parameters are indeed quite small , and furthermore they have very plausible values .
the time of hubble exit is typically of order 60 hubble times before the end of inflation , depending mainly on the reheat temperature , which means that the natural time scale of variation is of order @xmath194 .
if each time derivative in the right - hand expressions of eqs .
( [ ahg : eq14 ] ) and ( [ ahg : eq15 ] ) is replaced by a factor of @xmath195 , one sees that the slow roll parameters are plausibly of order 1/60 . the case @xmath196 is called scale - invariant , because it means that @xmath197 is independent of @xmath142 ; that is , each mode has the same strength , at the time of hubble exit , as any other mode .
since @xmath197 is constant while the wavelength is long compared to the hubble length , all modes also have the same strength at the time of hubble reentry .
single - field inflation produces density fluctuations that are approximately scale - invariant , because all the modes that are visible today passed through hubble exit during a small interval of time during inflation , so the conditions under which they were generated were very similar .
in addition to the nearly scale - invariant spectrum that we just calculated , there are two other key features of the density fluctuations that follow as a consequence of slow - roll single - field inflation .
the first is that the fluctuations are adiabatic , which means that every component of the matter in the universe the photons , the baryons , and the dark matter fluctuate together .
the temperature can be related to the density of photons , so it also fluctuates with the density baryons or dark matter .
the reason for this feature is clear , because the time delay affects all properties of the matter in the universe the same way .
until the perturbations reenter the hubble length ( after which complicated things can happen ) , every region of space behaves just like any other region of space , except for a time offset .
thus the matter content of any one region can differ from that of some other region by at most an adiabatic compression or expansion .
the wmap team @xcite has tested this relation for the possibility of non - adiabatic fluctuations between photons and cold dark matter . by combining wmap data with other data , they find at the 95% confidence level that the non - adiabatic component is at most 6% of the total in the case of `` axion - type '' perturbations , or 0.4% in the case of `` curvaton - type '' perturbations .
the other key predication of slow - roll single - field inflation is that the perturbations should be gaussian . why are they gaussian ?
they are gaussian because @xmath198 is calculated in a quantum field theory .
the perturbations are small so we expect accurate results at lowest order , which means that we are only calculating free - field - theory expectation values , and they are gaussian .
there are of course higher order corrections , which in a given model can also be calculated , but they are generically very small .
so , to first approximation , we expect the answers to be gaussian , which means in particular that the three - point correlation function should vanish .
there has been a lot of effort to look for non - gaussianity , but so far no convincing evidence for non - gaussianity has been found .
the calculations shown here stop just after the end of inflation , but with a lot of work by many astrophysicists the calculations have been extended to make detailed predictions for the fluctuations that can be detected today in the cosmic microwave background .
the success is beautiful . to process the data , the temperature pattern observed in the cmb
is expanded in spherical harmonics , which is the spherical equivalent of fourier transforming , providing information about how the intensity of the fluctuations varies with angular wavelength .
figure [ ahg : fig3 ] @xcite shows the observed temperature fluctuations as a function of the multipole number @xmath199 , using the 7-year wmap data @xcite for @xmath200 , and acbar data @xcite for higher @xmath199 .
the red line is the theoretical curve that comes about by extending the inflationary predictions to the present day in a model with dark energy ( @xmath201 ) and cold dark matter , using the best - fit parameters found by the wmap team @xcite : @xmath202 , @xmath203 , @xmath204 , @xmath205 , @xmath206 , and @xmath207 , where @xmath208 is the optical depth experienced by the photons since the `` recombination '' of the primordial plasma at about 380,000 years after the big bang .
while there are 6 free parameters , 4 of them have values that are expected on the basis of theory ( @xmath209 ) or other observations ( @xmath210 , @xmath211 , and @xmath212 ) , and they agree well .
one of the free parameters determines the overall height , so one should not be impressed that the height of the primary peak matches so well .
but the location , shape , and relative heights of the peaks are really being predicted by the theory , so i consider it a spectacular success . for comparison , the graph also shows predictions for several alternative theories , all of which are now ruled out by this data .
the yellow line shows the expected curve for an open universe , with @xmath213 .
the green line shows an inflationary model with @xmath214 , but with @xmath215 and no dark energy .
the magenta line shows the expectations for fluctuations generated by the formation of cosmic strings in the early universe , taken from ref . .
structure formation caused by cosmic strings or other `` defects '' was considered a viable possibility before this data existed , but now cosmic strings are completely ruled out as a major source of density fluctuations .
( there are possibly alternative ways to generate density perturbations with the same properties as those of inflation , but there is not yet a consensus about how easy it is to construct a plausible model .
the cyclic ekpyrotic model @xcite was claimed to naturally produce such fluctuations , but these claims were disputed by a number of authors @xcite .
now at least some of the founders of ekpyrosis @xcite agree that the original models do not give a nearly - scale invariant spectrum , as had been claimed .
but these papers and others have proposed newer , more sophisticated versions of bouncing universes , generally involving either multiple fields , or settling for scale invariance for only a limited range of scales .
baumann , senatore , and zaldarriaga @xcite have argued that any single - field model with attractor behavior has to be very close to de sitter space to remain weakly coupled for at least the required @xmath21610 @xmath35-folds needed to account for observations . )
there are still a number of important , outstanding questions concerning density perturbations : will @xmath217-modes be found ?
experiments are starting to measure the polarization of the cmb , for which the spherical harmonic expansion for the temperature pattern is replaced by an expansion in @xmath218-modes and @xmath217-modes @xcite .
the @xmath218-modes are those that can be expressed as gradients of scalar harmonic functions , and they are produced as a by - product of the density perturbations that we have been discussing .
the @xmath217-modes are orthogonal to the @xmath218-modes ; they can not be expressed as gradients of scalar modes , and they can not be produced by density perturbations .
there can be foreground contamination , but the only known primordial source of @xmath217-modes is a background of gravitational waves .
thus , gravity waves might be discovered in the cmb before they can be seen directly .
the discovery of a primordial gravity wave background would be very exciting , because it is the only thing that will give us a clue about the energy scale at which inflation happened .
as far as we know now inflation might have happened anywhere from the electroweak scale up to the grand unified theory ( gut ) scale , or a little beyond .
the discovery of gravity waves would end the uncertainty , and would also give strong evidence for the inflationary picture .
there are , however , many inflationary models for which the energy scale would be too low for the gravitational waves to be visible . can sub - planckian physics influence the calculation of inflationary density perturbations ?
a typical gut - scale inflationary model would include about 60 @xmath35-folds of inflation , expanding by a factor of @xmath219 . from the end of inflation to today the universe would expand by another factor of @xmath220 .
this means that a distance scale of 1 m today corresponds to a length of only about @xmath221 m at the start of inflation , 18 orders of magnitude smaller than the planck length ( @xmath222 m ) . with a little more than the minimal amount of inflation
which would be a certainty in the eternal inflation picture to be discussed below even the largest scales of the visible universe would have been sub - planckian at the start of inflation .
so , it is relevant to ask whether inflation can possibly offer us a glimpse of sub - planckian physics .
there is of course no solid answer to this question , since there is no real understanding of how this process should be described .
kaloper , kleban , lawrence , and shenker @xcite have argued that the perturbations are determined primarily by local effective field theory on the scale of order @xmath36 , so that sub - planckian effects would be invisible except possibly in unconventional models for which the fundamental string scale is many orders of magnitude below the four - dimensional planck mass , @xmath223 gev .
some authors @xcite have reached similar conclusions , but other authors @xcite have concluded that the effects might be much easier to see .
the conclusions of ref . seem plausible to me , but certainly the role of sub - planckian physics is not yet fully understood .
will effects beyond the single - field slow - roll approximation be found ? with multiple fields , or with unusual features in the potential for a single field , models can be constructed that predict significant non - gaussianity , non - adiabaticity , or spectral distortions .
there is an active industry engaged in studying models of this sort , and in looking for these nonstandard features in the data .
the wmap seven - year analysis @xcite reports `` no convincing deviations from the minimal model , '' but we all await the data from the planck mission , expected in less than a year , and the data from a variety of ground - based experiments .
since the density perturbation calculations have been incredibly successful , it seems to make sense to take seriously the assumptions behind these calculations , and follow them where they lead .
i have to admit that there is no clear consensus among cosmologists , but to many of us the assumptions seem to be pointing to eternal inflation , and the multiverse . the mechanism for eternal inflation is described most efficiently by separating the cases of the two types of potential functions shown in fig .
[ ahg : fig1 ] . for the new inflation case ,
that state for which the scalar field is poised on the top of the potential hill is a metastable state , often called a `` false vacuum , '' which decays by the scalar field rolling down the hill .
this state decays exponentially , but in any working model of inflation the half - life of the decay is much longer than the doubling time associated with the exponential expansion .
thus , if we follow a region for a period of one half - life , at the end of the period only half of the original region would be still be inflating .
however , the half that is still inflating will have a volume vastly larger than the volume of the entire region at the start , so the process will go on forever .
each decay will lead to the production of a `` pocket '' universe , and the creation of pocket universes will go on forever , as pieces of the ever - growing false vacuum region undergo decays .
once inflation starts , it never stops . for the case of a chaotic - type potential , as in fig .
[ ahg : fig1](b ) , naively one would think that the field would inexorably roll down the hill in some finite amount of time .
however , linde @xcite discovered that when quantum fluctuations are taken into account , this need not be the case . to understand this , consider an inflating region of space of size @xmath51 , with the inflaton field @xmath9 approximately uniform over this region , at some value @xmath103 .
after one hubble time ( @xmath51 ) the region will have expanded by @xmath224 , and can be viewed as 20 hubble - sized regions which will start to evolve independently .
the average field @xmath9 in any one of these regions will usually be lower than @xmath103 , due to the classical rolling down the hill , but the classical evolution will be modified by random quantum jumps , which can be estimated as @xmath225 .
it is therefore possible that in one or more of these 20 regions , @xmath9 can equal or exceed @xmath103 .
a back - of - the - envelope calculation shows that if @xmath226 then the expectation value for the number of regions with @xmath227 is greater than one .
that implies that the number of hubble - sized regions with @xmath227 will grow exponentially with time , and the inflation becomes eternal .
note that @xmath228 , so the eternally inflating behavior is really the large-@xmath9 , long - wavelength , tail of the density perturbation spectrum . since @xmath229 grows without bound as @xmath230 for most potentials under consideration ,
almost all models allow for eternal inflation .
there is certainly no proof that we live in a multiverse , but i will argue that there are three winds that is , three independent scientific developments , arising from three different branches of science which seem to be leading to the multiverse picture . _
string theory : the landscape .
_ string theory predicts that there is not just one kind of vacuum , but instead there are a colossal number of them : @xmath231 or maybe more @xcite .
the underlying laws of physics would be the same everywhere , but nonetheless each type of vacuum would create an environment in which the low - energy laws of physics would be different . thus ,
if there is a multiverse , it would be a varied multiverse , in which the different pocket universes would each appear to have their own laws of physics .
_ observational astronomy : the cosmological constant . _
the third `` wind '' has its roots in the fine - tuning that our universe appears to exhibit . in the past
a minority of physicists argued that things such as the properties of ice or the energy levels of carbon-12 appeared to be fine - tuned for the existence of life , but not very many scientists found this convincing .
if these properties were different , then maybe life would form some other way .
however , a form of fine - tuning that many of us find much more convincing became evident starting in 1998 , when two groups of astronomers @xcite announced that the expansion of the universe is not slowing down due to gravity , but is in fact accelerating .
the simplest explanation is that the acceleration is caused by a nonzero energy density of the vacuum , also known as a cosmological constant .
but that would mean that the vacuum energy density is nonzero , yet a full 120 orders of magnitude smaller than the planck scale ( @xmath232 , where @xmath233 ) , the scale that most theoretical physicists would consider natural .
physicists have struggled to find a physical explanation for this small vacuum energy density , but no generally accepted solution has been found .
but if the multiverse is real , the problem could go away .
with @xmath231 different types of vacuum , a small fraction , but nonetheless a large number of them , would be expected to have an energy density as small as what we observe .
the smallness of the vacuum energy density would be explained , therefore , if we could explain why we should find ourselves in such an unusual part of the multiverse .
but as pointed out by weinberg and collaborators @xcite some time ago , there is a selection effect .
if we assume that life requires the formation of galaxies , then one can argue that life in the multiverse would be concentrated in those pocket universes with vacuum energy densities in a narrow band about zero .
thus , while a typical vacuum energy density in the multiverse would be on the order of the planck scale , almost all life in the multiverse would find a small value , comparable to what we see . while the multiverse picture looks very plausible in the context of inflationary cosmology at least to me it raises a thorny and unsolved problem , known as the `` measure problem . ''
specifically , we do not know how to define probabilities in the multiverse . if the multiverse picture is right , then anything that can happen will happen an infinite number of times , so any distinction between common and rare events requires the comparison of infinities .
such comparisons are not mathematically well - defined , so we must adopt a recipe , or `` measure , '' to define them . since the advent of quantum theory essentially all physical predictions have been probabilistic , so probability is not a concept that we can dispense with . to date
we do not understand the underlying physical basis for such a measure , but much progress has been made in examining proposals and ruling out many of them .
one might guess that this problem is easily handled by choosing a finite sample spacetime region in the multiverse , calculating the relative frequencies of different types of occurrences in the sample region , and then taking the limit as the region becomes infinite .
this seems like a very reasonable approach , and in fact most of the measure proposals that have been discussed are formulated in this way .
the problem is that the answers one obtains are found to depend sensitively on the method that is used to choose the sample region and to allow it to grow .
the dependence on the method of sampling seems surprising , but it sounds plausible if we remember that the volume of the multiverse grows exponentially with time .
a sample spacetime region will generally have some final time cutoff , and the spacetime volume will generally grow exponentially with the cutoff .
but then , no matter how large the cutoff is taken , the volume will always be dominated by a region that is within a few time constants of the final time cutoff hypersurface .
no matter how large the final time cutoff is taken , the statistics will never be dominated by the interior of the sample region , but instead instead will be dominated by the final time cutoff surface . for that reason , it is not surprising that the method of choosing this surface will always affect the answers .
there are a number of important questions , in the multiverse picture , that depend very crucially on the choice of measure .
how likely is it that we observe a vacuum energy density as low as what we see ?
how likely is it that our universe has a mass density parameter @xmath234 sufficiently different from 1 that we can hope to measure the difference ?
how likely is it that we might find evidence that our pocket universe collided with another sometime in its history ? and , if there are many vacua in the landscape of string theory with low energy physics consistent with what we have measured so far , how likely is it that we will find ourselves in any particular one of them ?
if we live in a multiverse , then in principle all probabilities would have to be understood in the context of the multiverse , but it seems reasonable to expect that any acceptable measure would have to agree to good accuracy with calculations that we already know to be successful . as discussed in ref .
, we can identify a class of measures that give reasonable answers .
it seems plausible that the ultimate solution to this problem will give similar answers , but the underlying principles that might determine the right answer to this question remain very mysterious .
nonetheless , the success of inflation in explaining the observed properties of the universe , including the density perturbation predictions discussed here , provides strong motivation to expect that some solution to the measure problem will be found .
j. khoury , b. a. ovrut , p. j. steinhardt and n. turok , , 123522 ( 2001 ) [ hep - th/0103239 ] .
j. khoury , b. a. ovrut , p. j. steinhardt and n. turok , , 046005 ( 2002 ) [ hep - th/0109050 ] .
a. j. tolley , n. turok and p. j. steinhardt , , 106005 ( 2004 ) [ hep - th/0306109 ] .
j. khoury , p. j. steinhardt and n. turok , , 161301 ( 2003 ) [ astro - ph/0302012 ] .
j. khoury , p. j. steinhardt and n. turok , , 031302 ( 2004 ) [ hep - th/0307132 ] .
r. brandenberger and f. finelli , , 056 ( 2001 ) [ hep - th/0109004 ] .
j. steinhardt , in _ the very early universe , _ proceedings of the nuffield workshop , cambridge , 21 june 9 july , 1982 , eds : g. w. gibbons , s. w. hawking , and s. t. c. siklos ( cambridge university press , cambridge , 1983 ) , pp . | this article discusses density perturbations in inflationary models , offering a pedagogical description of how these perturbations are generated by quantum fluctuations in the early universe .
a key feature of inflation is that that rapid expansion can stretch microscopic fluctuations to cosmological proportions .
i discuss also another important conseqence of quantum fluctuations : the fact that almost all inflationary models become eternal , so that once inflation starts , it never stops . |
here we outline the calculation of the coulomb couplings between the two - electron orbital states associated with each qubit .
since all the gate operations considered in this work are performed by tuning the bias between the two qds , the hund - mulliken orbital hamiltonian must account for biased dot configuration .
more details are given in @xcite .
we start by approximating the orbitals for the biased double - dot by those of two harmonic wells centered at @xmath137 , where @xmath138 is the interdot half separation , @xmath139 is the qd bohr radius , and @xmath140 is a dimensionless parameter corresponding to the interdot bias ( normalized to the qd confinement ) .
we denote the orthonormalized ( predominantly ) left and right single particle orbitals by @xmath141 , and use them to construct the two - electron spin basis states .
these are the two doubly occupied singlet states , @xmath142 , @xmath143 , and the separated singlet and unpolarized triplet states , @xmath144 , @xmath145 , where@xmath146 indicates the number of electrons in each dot .
since the orbital hamiltonian does not connect the triplet state with any of the singlet states , we only need to diagonalize the @xmath147 singlet block , resulting in a hybridized singlet state .
we note that both the polarized and doubly occupied triplet states are neglected in our analysis .
doubly occupied triplet states require one of the electrons to reside in an excited orbital , thus their energy is far above the doubly occupied singlets @xcite . utilizing overhauser fields to generate a magnetic field gradient
will likely create a small perpendicular component that may mix the @xmath148 computational subspace with the @xmath149 subspace , spanned by the polarized triplet states . with @xmath150 , the transverse inhomogeneous field should not exceed @xmath151 nev for which leakage errors are found to be below @xmath152 @xcite , thus polarized triplets can be safely neglected as well .
for the calculation of the inter - qubit coulomb couplings , it is convenient to work in the orthonormal basis of symmetric and antisymmetric orbital combinations : @xmath153 , @xmath154 . in this basis ,
the coulomb interaction operator between the two qubits is written as @xmath155 where @xmath156 is the electron charge density operator for either qubit , and @xmath157 denote the qubit orbital state ( symmetric or antisymmetric combination ) .
the coulomb matrix elements in eq .
( [ bg ] ) in the main text are found in terms of the @xmath158 operators : @xmath159 \nonumber \\
v_{s t_0 } \!&\!=\!&\ !
v_{t_0 t_0 } + { \cal n}_t^2 \left ( a_1^t + a_2^t \right)\left [ \sqrt{2 } \left ( f_{1111}+f_{1122 } \right .
\nonumber \\ \!&\!-\!&\ ! \left .
f_{2211}-f_{2222 } \right)+\left(a_1^t - a_2^t\right)\left ( f_{1211}+ f_{1222 } \right .
\!&\!+\!&\ ! \left .
f_{2111}+f_{2122 } \right ) \right]\\ v_{ss } \!&\!=\!&\ !
s}+v_{s t_0}+{\cal n}_t^2 { \cal n}_c^2 \left ( a_1^t+a_2^t \right ) \nonumber \\ \!&\ ! \times \!&\ !
\left ( a_1^c+a_2^c \right)\left [ 2\left ( { f_{1111}+f_{2222}-f_{1122}-f_{2211 } } \right ) \right
. \nonumber \\
\!&\!+\!&\ ! \left .
\sqrt 2 \left ( a_1^t - a_2^t \right)\left ( f_{1211}+f_{2111}-f_{1222}-f_{2122 } \right ) \right . \nonumber \\
\!&\!+\!&\ ! \left .
\sqrt 2 \left(a_1^c - a_2^c \right)\left ( f_{1112}+f_{1121}-f_{2212}-f_{2221 } \right ) \right .
\!&\!+\!&\ ! \left .
\left(a_1^t - a_2^t\right)\left(a_1^c - a_2^c \right ) \right .
\nonumber \\ \!&\!\times \!&\ !
\left.\left ( f_{1212}+ f_{2121}+f_{1221}+f_{2121 } \right ) \right ] , \nonumber\end{aligned}\ ] ] where the superscript @xmath14 ( @xmath15 ) denotes target ( control ) qubit , @xmath160 are the @xmath161 , @xmath162 components of the lowest lying hybridized singlet , respectively , and @xmath163 is the normalization .
the coulomb interaction terms @xmath158 are calculated by evaluating the electrostatic energy associated with placing the control qubit charge distribution in the potential @xmath164 , that is due to the target qubit charge distribution . denoting the charge , dipole , and quadrupole electric moments associated with the qubit charge distribution by @xmath165 , respectively , we obtain @xmath158 up to and including quadrupole - quadrupole order : @xmath166 \!. \nonumber\end{aligned}\ ] ] in deriving eq .
( [ multipole ] ) we assumed both qubits are aligned with the @xmath167-axis .
the qubit dipole moments are thus in the @xmath167 direction for both qubits , @xmath168 .
the quadrupole moments , @xmath169 , are found to have only diagonal elements ( @xmath170 ) .
explicit expressions for the various multipole moments are given in @xcite , and are used to obtain the inter - qubit coupling terms , to quadrupole - quadrupole order with the following nonvanishing contributions : @xmath171 the first ( second ) superscript in each term denotes contribution from the particular multipole moment : monopole ( 1 ) , dipole ( 2 ) , and quadrupole ( 4 ) of the target ( control ) qubit charge distribution . the effective exchange contribution for the control qubit , @xmath172 ,
is found from eq .
( [ beta ] ) by interchanging the left and right superscripts .
symmetry considerations suggest that for identical qubits , equally biased , one has : @xmath173 , and @xmath174 .
note that the dipole moment of a non - biased double dot vanishes , therefore there are no contributions to @xmath175 that involve a non - biased target qubit dipole moment .
similarly @xmath172 will have no control qubit dipole contributions for non - biased control qubit , and @xmath33 will have no dipole contributions from either qubits if both are non - biased .
figure [ figs1 ] shows the dependence of the various multipole terms on the inter - qubit distance for the case of two double - dots equally biased at the singlet anticrossing point .
the figure demonstrates the convergence of the multipole expansion in the considered distance range .
we conclude that for @xmath176 nm , used throughout most of this work , both @xmath28 and @xmath33 couplings are well approximated by their respective leading contributions . at the sweet - spot one
has : @xmath180 , @xmath77 .
we consider dephasing of the target qubit , initially prepared in the state : @xmath181 , by solving for the time evolution of the system of two qubits under the hamiltonian , eq .
( [ h ] ) , and tracing out the control qubit states .
notice that the term dephasing is used here and in the main text to quantify different time evolutions of @xmath0 and @xmath1 constituents in a superposition state due to inter - qubit coupling .
this is not pure dephasing in the sense of loss of coherence due to entanglement with a bath . for a control field gradient satisfying @xmath182 , which is easily met at the low - bias sweet - spot with a conservative magnetic field gradient of @xmath183 mt , we find : @xmath184 while oscillations are much more rapid ( @xmath185 ) than in the @xmath94 case ( @xmath186 ) , their amplitude is very small , thus the qubits are effectively decoupled . to gain better insight to the qubit decoupling bias points and their robustness against bias errors we plot in figs .
[ figs2](a , b ) target qubit dephasing times vs. the control qubit bias , measured from the @xmath187 , low - bias position .
we define dephasing time as the time it takes @xmath188 to drop to 50% of its original value of @xmath189 , due to coupling with the control qubit . in general , we note that the effects of the control qubit on target dephasing are minimal at a relatively wide control bias region around the sweet spot .
this should be contrasted with a much higher sensitivity to bias errors at the high - bias position . when @xmath190 , target dephasing at the sweet spot is identical for both control states , but fig .
[ figs2](a ) shows that slight detuning of the control bias results in complete qubit decoupling when the control is in either singlet or triplet state . to explain this behavior we notice that at bias positions satisfying @xmath191 in both qubits , eq .
( [ h ] ) leads to : @xmath192 for the plus sign , and @xmath193 for the minus sign . as a result ,
the qubits are decoupled when the control is in a singlet ( triplet ) state for the plus ( minus ) sign , and the target qubit dephases as @xmath194 when the control is in the other state .
since @xmath172 grows faster than @xmath195 with @xmath21 , the intersection @xmath196 occurs at a negative bias , and similarly @xmath197 occurs at a positive bias relative to the sweet spot position ( see figure [ fig1](b ) ) .
these considerations are numerically confirmed by correcting the sweet spot to satisfy @xmath198 for @xmath77 . fig .
[ figs2](c ) shows qubit decoupling at this @xmath33-corrected sweet spot position for singlet control state .
when @xmath182 , the qubits decouple at the original sweet spot ( with errors of order @xmath199 ) according to eq .
( [ rho12b ] ) , for both control states , thus shifting the sweet spot bias results in a similar shift in the decoupling bias position , as seen in fig .
[ figs2](d ) .
finally , errors in @xmath6-rotations performed at the @xmath49 bias position can be evaluated , using eq .
( [ h ] ) . considering a @xmath46-flip on a target qubit ,
initially prepared in a singlet state , we take @xmath200 and find the leading contribution in a @xmath201 expansion . at the end of the pulse , the gate error , given by the singlet probability is found to be : @xmath202 irrespective of the control qubit state . when @xmath203 eq .
( [ xerr ] ) reduces to @xmath204 , given in the main text .
notice , that when @xmath205 , where @xmath206 is an odd integer larger than 1 , the leading contribution , eq .
( [ xerr ] ) , vanishes and we get @xmath6-gate error of the order @xmath207 .
here we use a variant of the bias sequence generating @xmath6-rotations that was proposed in @xcite .
our cycle consists of two working points : ( i ) the low - bias symmetric sweet - spot @xmath208 , where rotations are about the @xmath6 axis , and ( ii ) a bias position slightly above the @xmath208 point , where @xmath209 and the rotation is about a @xmath58-tilted axis where @xmath210 .
when @xmath64 , a @xmath5-rotation by an arbitrary angle @xmath62 can be generated by the cycle : @xmath211 where @xmath57 and @xmath212 are rotation matrices about the @xmath58-tilted and @xmath6 axes , respectively , given by : @xmath213 the angles @xmath60 and @xmath61 are functions of @xmath58 and @xmath62 , and are found to be similar ( but not identical ) to those given in @xcite : @xmath214 where the plus ( minus ) sign in eq .
( [ phi ] ) corresponds to positive ( negative ) @xmath58 . in particular , a @xmath46-rotation about the @xmath5 axis is obtained with : @xmath215 we consider a @xmath46-rotation about the @xmath5 axis by taking the second working point to satisfy @xmath65 so that the bias sweep from the low - bias sweet - spot is minimized .
the pulse durations are set to : @xmath216 , and @xmath217 , at the @xmath58-tilted point and low - bias sweet - spot , respectively .
to simulate this cycle we discretize bias and time steps , so that the actual switching time between the working points is taken into account .
bias switching should be nonadiabatic with respect to nuclear mixing , but slow as compared with the tunnel splitting of the hybridized singlet states . bias sweep rates @xmath218 mv
/ ns , comparable to experimental values @xcite , do not contribute appreciably to gate errors presented in this work , and can be further corrected by pulse design .
[ figs3](a ) depicts the target off - diagonal density matrix element during a @xmath46-rotation about the @xmath5 axis from @xmath219 , using the sequence in eq .
( [ txt ] ) .
taking @xmath220 we use eq .
( [ h ] ) to find the deviation of the the target off diagonal density matrix element from @xmath221 at the end of the cycle .
the leading error terms in a @xmath222 , @xmath71 expansion , where @xmath223 ( @xmath69 ) is evaluated at the sweet - spot ( @xmath58-tilted ) position , are found to be : @xmath224\ !
, \label{zerr}\end{aligned}\ ] ] and are compared with the numerical simulations in fig .
[ figs3](b ) . the proposed sequence , operating around the low - bias sweet - spot ,
is shown to generate high fidelity @xmath5-rotations at a reasonably short gate time . for demonstration purpose
an effective cnot gate for an initial target qubit in either @xmath87 or @xmath81 can be implemented by biasing both qubits to the point @xmath225 , @xmath77 , close to the high - bias symmetric sweet - spot discussed in the main text . at this bias ,
the diagonal part of the two - qubit hamiltonian is @xmath226 , and @xmath227 .
thus , the target state will effectively rotate about the @xmath5-axis when the control is in @xmath228 and about the @xmath6-axis when the control is in @xmath81 . applying this pulse for @xmath229 to a two - qubit system , with the target initially in either of the computational states ( but not in a superposition of the two ) , results in a cnot gate , with gate time inversely proportional to @xmath39 .
the degeneracy in the hamiltonian at the above bias point induces gate errors that scale linearly with @xmath230 , and therefore are rather large for @xmath20 nm ( @xmath231 ) .
a better bias position is the slightly asymmetric point : @xmath232 , @xmath233 , at which the diagonal part of the two - qubit hamiltonian is @xmath234 . at this position gate errors
are found to scale as @xmath235 , and are small over a wide range of @xmath39 values .
cnot gate can also be implemented using the scheme proposed in @xcite , but this scheme requires biasing the two double dots in opposite directions , so that the configuration : @xmath236 is obtained .
this bias position is very far from the low - bias symmetric point that is proposed in the current work for single qubit operations , and could be experimentally challenging .
here we demonstrate the effect of the device size ( and hence the number of charge traps ) on qubit dephasing and sample - to - sample variability .
[ figs4 ] ( a ) and ( b ) show calculated fid and se dephasing , respectively , for a device of area @xmath237
( @xmath129 , @xmath130 ) , considered in the main text .
the solid blue lines depict average decay over 10,000 random tlf sets , and the dashed green ( red ) lines correspond to maximally ( minimally ) affecting sets .
the results show a large sample - to - sample variability , in particular for the se decay , which is governed by the wide switching rate distribution , 1 ms@xmath2381 s. figs .
[ figs4 ] ( c ) and ( d ) show fid and se dephasing , respectively , for a device of area @xmath239
( @xmath240 , @xmath241 ) .
the fid dephasing time , @xmath242 ns , and variability are hardly changed with respect to the smaller device . in contrast , the se decay shows an order - of - magnitude reduction in dephasing time , @xmath243 , and much more reproducible decay , suggesting that for @xmath244 results become self averaging .
we note that in both qubit manipulation protocols considered here , the noise generated by the charge environment is non - gaussian , and therefore can not be described in terms of its spectral function alone . for larger devices , where the number of fluctuators is of the order of 100
, the distribution of coupling strengths and switching rates can be averaged , and the resulting noise is dominated by a large number of weakly coupled tlfs , demonstrating characteristics of @xmath245 spectrum . the dependence of qubit dephasing on tlf parameter distributions , qubit characteristics , and magnetic field , and its scaling with the number of qubits , as well as qubit dephasing under multiple - pulse dynamical decoupling sequences , will be presented elsewhere @xcite . | exchange - coupled singlet - triplet spin qubits in two gate - defined double quantum dots are considered theoretically . using charge density operators to describe the double - dot orbital states
, we calculate the coulomb couplings between the qubits , and identify optimal bias points for single- and two - qubit operations , as well as convenient idle positions .
the same intuitive formulation is used to derive dephasing rates of these qubits due to the fluctuating charge environment , thereby providing the main considerations for a quantum computation architecture that is within current experimental capabilities .
electron spins localized in semiconductor quantum dots ( qds ) have been vigorously pursued in recent years as potential qubit candidates , due to their relative isolation from the environment @xcite .
the long coherence times measured for single electron spins render their manipulation and readout particularly challenging .
this has prompted a number of proposals to encode the logical qubit into two - spin singlet ( @xmath0 ) and unpolarized triplet ( @xmath1 ) states @xcite .
gaas gate - defined lateral qds in particular have served as a useful platform to study quantum control and decoherence of exchange - coupled spin qubits .
the main reason is the high tunability of the two - electron exchange interaction , @xmath2 , that is controlled on a subnanosecond scale by changing the bias between the two dots . in the context of the @xmath3 qubit
this tunability enables initialization of the qubit state in a doubly occupied singlet state using positive interdot bias , where triplet states are energetically inaccessible . by applying a perpendicular magnetic field ,
the two polarized triplet states are zeeman splitted , further reducing leakages from the logical basis states .
recent experiments in gated double dots have implemented spin - echo @xcite and pulse optimization techniques that have pushed two - spin qubit dephasing time over @xmath4 @xcite . in this work
we consider two proximal @xmath3 qubits , each formed by two electrons localized in a gate - defined double dot , as depicted in fig .
[ fig1](a ) @xcite .
single qubit @xmath5-rotations are fixed by pulsing the interdot bias that controls @xmath2 , whereas @xmath6-rotations are generated by a magnetic field gradient across the two dots . a promising method to produce
such gradient is to utilize the hyperfine coupling at the @xmath7 degeneracy point to dynamically induce unequal nuclear polarization across the two dots . implementing these electrically controlled pump cycles , foletti _ et al .
_ were recently able to sustain a stable overhauser field gradient of @xmath8ev @xcite .
the single - qubit hamiltonian can thus be written as @xmath9 , where @xmath10 , and @xmath11 is the vector of pauli spin matrices for the pseudospin states @xmath0 , and @xmath1 .
when two double dots are placed near each other , their respective electrons become coulomb coupled .
this coupling depends on the orbital states and thus on the spin configurations of the electrons , resulting in an effective additional exchange interaction for each qubit , and spin dependent entangling coupling .
below we show that for certain bias points the first term turns off the exchange interaction that is otherwise constantly on .
the latter term provides a controlled - phase ( cphase ) gate , thus completing a universal set of gates that is based solely on accurate control over the interdot bias at each qubit .
in particular , the requirement to control the tunnel coupling between the dots , needed to tune @xmath2 to zero @xcite , and the need to perform single - qubit @xmath5-rotations near the dephasing - susceptible avoided crossing point @xcite , are eliminated , and qubit decoupling positions are identified .
_ inter - qubit coulomb couplings and sweet spots .
_ the coulomb interaction between two qubits , each formed by the singlet and triplet spin states of two electrons in a biased double dot , is found by considering the charge distributions of the electron orbital states , assuming no inter - qubit tunnel coupling .
up to a constant term , the interaction hamiltonian can be written in the form @xcite : @xmath12 where @xmath13 here the subscript @xmath14 ( @xmath15 ) denotes target ( control ) qubit , and @xmath16 are the coulomb matrix elements involving the hybridized singlet ( @xmath0 ) and unpolarized triplet ( @xmath1 ) states , where the left ( right ) subscript corresponds the target ( control ) qubit .
for nominally identical , equally biased qds , we have @xmath17 . when the inter - qubit distance @xmath18 is sufficiently large , there is no tunneling between the qubits , the two charge distributions are separated in space , and the coulomb interaction between them can be described systematically using a multipole expansion approach @xcite .
for the same - axis geometry shown in fig .
[ fig1](a ) , couplings are always positive , and they are plotted in fig . [ fig1](b ) as a function of the target qubit bias , @xmath19 , for @xmath20 nm and a fixed control qubit bias , @xmath21 , corresponding to the anticrossing point . here and for all results reported in this work we have considered @xmath22 mt ( @xmath23 ) , and have modeled the double dots using a quartic potential with dot confinement @xmath24 ( @xmath25 nm ) , and half interdot distance @xmath26 . while all the coulomb couplings depend on both @xmath27 , for the parameter range considered ,
each of the @xmath28 couplings is predominantly controlled by its respective qubit bias ( as seen by the red dotted line in fig . [ fig1](b ) ) .
= 0.95 complementing the target and control single - qubit hamiltonians with the interaction terms , eq .
( [ hint ] ) , we find that at certain bias points exchange is turned off ( gray circles in fig .
[ fig1]b ) .
we define a sweet spot in the 2d bias space @xmath27 as a point that satisfies two conditions : @xmath29 , and @xmath30 .
the first condition facilitates an idle position of the target qubit , or a pure @xmath6-rotation in the presence of a magnetic field gradient , @xmath31 .
similarly , the second condition is set to turn off the control qubit exchange , resulting in minimal interference with the target operation .
for @xmath32 nm we find that for each qubit there are always two such intersections , thus in the 2d bias space there are four possible working positions : two symmetric low- and high - bias points , and two asymmetric points .
as explained below , the symmetric low- and high - bias points are optimal for single- and two - qubit operations , respectively .
fig .
[ fig2](a ) depicts the dependence of the symmetric low - bias ( solid red line ) and high - bias ( solid blue line ) sweet - spot positions on the inter - qubit distance @xmath18 .
the corresponding entangling coupling , @xmath33 , at these bias positions is shown in fig .
[ fig2](b ) . for almost the entire range ,
the low - bias @xmath33 values exhibit a counter - intuitive increase with @xmath18 due to the increased sweet - spot bias as @xmath18 grows .
the strong bias dependence of @xmath33 overtakes its expected decrease with @xmath18 for the considered range .
= 0.7 _ idle position and single - qubit gates .
_ while the @xmath33 coupling can not induce relaxation of the two - electron qubit states , it generates dephasing of a superposition @xmath0-@xmath1 state , and gate errors . inspecting the two - qubit hamiltonian , we find that without magnetic field gradients , @xmath34 , when the two qubits are biased at a sweet - spot , preparing one of them in the state @xmath35 results in a time evolution of its off - diagonal density matrix element : @xmath36 , irrespective of the state of the other .
choosing the low - bias sweet - spot and @xmath20 nm , we have @xmath37 nev , corresponding to dephasing time of @xmath38s . since nuclear state preparation that generates a stable magnetic field
gradient requires pumping times between 60 ms - 1 s @xcite , it may well be necessary to operate under a fixed @xmath39 in both qubits .
as long as @xmath40 , the oscillation amplitude of the off diagonal element of the target qubit density matrix reads @xmath41 .
taking a conservative value of @xmath42ev , corresponding to a field gradient of @xmath43 mt , we find that at @xmath20 nm , target qubit errors are below @xmath44 .
working with a finite @xmath45 , the sweet - spot is no longer an idle position for the target qubit , and the gate time for a @xmath46-flip @xmath6-rotation is 2 ns , with a leading gate error of @xmath47 , for @xmath48 @xcite .
if the @xmath49 point is used as an idle position , @xmath6-rotations can be erased either by waiting integer number of full periods , @xmath50 , or by applying the pulse sequence : @xmath51 , where @xmath52 is the acquired @xmath6-rotation during the wait time @xmath53 . to generate @xmath5-rotations with a finite @xmath39 one can positively bias the target qubit so that @xmath54 produces a nearly perfect @xmath5-rotation .
note , however , that to maintain short gate times , @xmath39 can not be too small , thus a large excursion from the sweet - spot is necessary to obtain sufficiently large @xmath55 . at this position , near the singlet anticrossing ,
the @xmath33 coupling increases dramatically , generating substantially larger gate errors .
in addition , a positively biased qubit may decohere faster due to environmental charge fluctuations . to stay near the @xmath49 point
, one can perform a three pulse sequence : @xmath56 , where @xmath57 is a rotation about the @xmath58-tilted axis , @xmath59 , and the rotation angles @xmath60 and @xmath61 are fixed by @xmath62 and @xmath58 @xcite .
a @xmath5-rotation by any angle @xmath62 can be performed using this cycle , as long as @xmath63 ( @xmath64 ) .
choosing @xmath65 to stay as close as possible to the @xmath49 point , we complete a three - pulse @xmath5-rotation of @xmath66 in 5 ns , with a gate error smaller than @xmath67 .
the dominant contribution to the @xmath5-gate error reads : @xmath68 , where @xmath69 is evaluated at the @xmath58-tilted bias position .
although @xmath69 increases with @xmath39 due to the necessary larger bias sweep to the @xmath70 position , the ratio @xmath71 is monotonically decreasing in the considered range , keeping the gate error very small for @xmath72 @xcite .
_ two - qubit gates .
_ we now consider the high bias symmetric point , where the qubit coupling @xmath33 is large : @xmath73 . without magnetic field gradients ,
the two - qubit hamiltonian is diagonal in the computational basis state @xmath74 , and reads at the sweet - spot : @xmath75 .
a cphase gate is conveniently obtained by tuning the bias in both double dots to satisfy @xmath76 , @xmath77 , at which @xmath78 . letting the system evolve for time @xmath79 ,
the resulting gate , up to a common phase , is : @xmath80 .
the position of the @xmath33-corrected high - bias symmetric point is shown by the dashed line in fig .
[ fig2](a ) . working with finite field gradients in both qubits
, we find three different gate error scalings : ( i ) population errors with control in @xmath81 @xmath82 , ( ii ) phase errors @xmath83 , and ( iii ) population errors with control in @xmath84 @xmath85 , the latter being negligible in the considered @xmath39 range .
we have performed numerical simulations of the time evolution of the two qubits under a controlled @xmath46-pulse , where the target qubit is initially prepared in the state : @xmath86 , and the control qubit is in either @xmath87 ( target remains in its initial state ) or @xmath81 ( target evolves to @xmath88 ) .
fig .
[ fig3](a ) shows the resulting cphase gate errors for @xmath20 nm ( @xmath89 ) , which follow closely the above scalings up to @xmath72 .
the small difference in the two phase errors with different control qubit states is due to the hamiltonian structure at the @xmath33-corrected working point . to construct a cnot gate , a hadamard gate , @xmath90 ,
should be applied to the target qubit before and after the controlled operation @xcite .
@xmath91 can be implemented by biasing the target qubit to a position adjacent to the low - bias sweet spot , where @xmath92 , for a duration @xmath93 .
we note that the control qubit should not rotate prior to the cphase gate .
while it may be possible to employ a pulse sequence to correct such spurious control rotation , here we take @xmath94 during the first hadamard target gate .
this may be implemented by , e.g. , delaying the nuclear state preparation in the control qubit by @xmath95 .
this pulse sequence was simulated by discretizing bias and time steps , so that the actual switching times between different bias points are taken into account .
fig .
[ fig3](b ) shows the resulting cnot gate errors vs. @xmath31 for @xmath20 nm . population and phase errors interchange with respect to their role in the cphase gate , therefore the phase error with control in @xmath96 is the largest here , scaling linearly with @xmath97 .
the results demonstrate increased errors for both small @xmath31 ( during the hadamard gates ) , and large @xmath31 ( during the cphase gate ) .
a minimum phase error of @xmath98 is obtained for @xmath99 , but with such a small gradient , single qubit gate times are on the order of 100 ns ( see fig .
[ fig3](c ) ) .
phase errors smaller than @xmath100 are obtained with @xmath101 .
other errors , and in particular singlet return errors , are considerably smaller .
= 0.95 we stress that the opposing requirements for small single- and two - qubit gate errors , while operating under fixed @xmath39 , are met by utilizing the high tunability of the @xmath33 coupling , whose values show more than six orders - of - magnitude difference between the low- and high - bias sweet - spots for @xmath20 nm . taking
@xmath102 nm will increase the @xmath33 ratio by another order - of - magnitude , resulting in a minimum phase error of @xmath103 at @xmath104 , and a wider range for which errors are below @xmath100 up to @xmath105 . reducing the inter - qubit distance may , however , induce errors due to electron exchange coupling between the two double dots that are not included in the current analysis .
further optimization of gate errors and times is likely to be obtained by engineering the dot size and tunnel coupling within each double dot @xcite .
_ qubit dephasing due to fluctuating charge environment . _
exchange - coupled spin qubits are vulnerable to dephasing induced by charge noise , due to the different charge distributions of the singlet and triplet states @xcite .
the formulation we employed to find inter - qubit couplings can be used to quantify couplings of each of the qubits to nearby two - level charge fluctuators ( tlfs ) and evaluate the resulting qubit dephasing .
we identify two types of tlfs : ( i ) @xmath28-coupled , where the charge jumps in and out of the trap , and ( ii ) @xmath33-coupled , where the charge fluctuates between two sites in the trap .
while @xmath28-coupled tlfs have stronger coupling to the spin qubits , they are less abundant and locate farther from the qds , near the 2deg layer or quantum point contacts . starting with a single tlf
, we represent the qubit state with a bloch vector , and solve for its precession under the pseudo - field : @xmath106 @xcite .
@xmath107 represents a poisson switching process , and @xmath108 is the qubit - tlf coupling strength for @xmath28-type and @xmath33-type tlfs , respectively . averaging over the stochastic process , the qubit signal dephasing is given by : @xmath109 , with @xmath110 .
for free induction decay ( fid ) , we find @xcite : @xmath111 e^{\pm\gamma \mu t } , \label{fid}\ ] ] where @xmath112 , @xmath113 is the average switching rate , @xmath114 , and @xmath115 is the tlf level splitting . to calculate signal decay during a general dynamical decoupling pulse sequence we write @xmath116 , where @xmath117 is a filter function in the time domain @xcite . for the spin echo ( se )
protocol , @xmath118 , where @xmath119 is the step function , and the signal decay is found to be : @xmath120 , \label{se}\end{aligned}\ ] ] where @xmath121 ( @xmath122 ) is the real ( imaginary ) part of @xmath123 . in order to estimate two - spin qubit dephasing due to an ensemble of tlfs , we use @xmath124 , where @xmath125 ( @xmath126 ) is the number of @xmath28-type ( @xmath33-type ) traps .
we estimate the trap density as @xmath127 , and the two double - dot device area as @xmath128 , resulting in an active number of traps : @xmath129 , @xmath130 @xcite . in this estimate
we assume that an area of radius @xmath131 nm around each qubit is depleted , so that traps in this area are not charge - active , setting up the maximum qubit - tlf coupling strength .
trap parameters include level splitting of @xmath132 , center radius of 5 nm , and intersite distance of 20 nm , chosen to characterize @xmath133-doped dopants in the insulator .
finally , trap switching rates are taken between 1ms @xmath134 s @xcite .
we have used eqs .
( [ fid ] ) and ( [ se ] ) to calculate dephasing of a qubit at the low - bias sweet spot , by averaging over 10,000 sets of tlfs with random locations , orientations , and switching rates , within the parameter ranges specified above . in the fid case we obtain dephasing time of @xmath135 ns , comparable to measured coherence times @xcite . for the se protocol
we find @xmath136 , two - orders - of - magnitude longer than measured se coherence times @xcite , attributed to nuclear spin fluctuations , mediated by the hyperfine interaction @xcite . in the quasi - static regime considered here , fid dephasing times are dominated by the strongest fluctuator(s ) , and are thus mostly sensitive to qubit bias and qubit - tls distance .
se dephasing times are dominated by the fastest fluctuator(s ) , and are sensitive to the maximal switching rate @xcite .
we stress that for such a mesoscopic system , these results are not self averaging and may vary considerably between samples . for se experiments with larger devices , sample - to - sample variability is reduced and dephasing times become comparable with those induced by the nuclei @xcite , thus both charge and nuclear environments should be considered in future device design and operation . in conclusion , we have quantified the couplings between two @xmath3 qubits , identifying optimal working points .
our proposed scheme relies solely on interdot bias tuning at each qubit , and offers several advantages , including high fidelity single- and two - qubit gates under fixed field gradients , and idle positions at which the qubits are effectively decoupled .
in addition , we have used our formulation to quantify qubit dephasing due to charge fluctuations .
the author thanks xuedong hu and hendrik bluhm for helpful discussions and acknowledges funding from research corporation .
99 r. hanson _ et al .
_ , rev .
mod .
phys . * 79 * , 1217 ( 2007 ) .
d. a. lidar , i. l. chuang , and k. b. whaley , phys .
rev .
lett .
* 81 * , 2594 ( 1998 ) . j. levy , phys .
rev .
lett . * 89 * , 147902 ( 2002 ) .
j. m. taylor _ et al .
_ , nature physics * 1 * , 177 ( 2005 )
. j. r. petta _ et al .
_ , science * 309 * , 2180 ( 2005 ) .
c. barthel _ et al .
_ , phys .
rev .
lett . * 105 * , 266808 ( 2010 ) .
h. bluhm _ et al .
_ , nature physics * 7 * , 109 ( 2011 ) .
d. stepanenko and g. burkard , phys .
rev .
b * 75 * , 085324 ( 2007 ) .
s. foletti _ et al .
_ , nature physics * 5 * , 903 ( 2009 ) .
h. bluhm _ et al .
_ , phys .
rev .
lett .
* 105 * , 216803 ( 2010 ) . fast control over the tunnel coupling in gate defined double dots is challenging and has not been demonstrated yet .
in other qd systems such as self - assembled or vertically stacked qds tunnel coupling is fixed by fabrication .
r. hanson and g. burkard , phys .
rev .
lett . * 98 * , 050502 ( 2007 ) .
guy ramon and xuedong hu , phys .
rev .
b * 81 * , 045304 ( 2010 ) .
see supplemental material at [ url will be inserted by publisher ] .
alternative implementations of cnot gate are briefly discussed in @xcite .
xuedong hu and s. das sarma , phys .
rev .
lett . * 96 * , 100501 ( 2006 ) .
dimitrie culcer , xuedong hu , and s. das sarma , appl .
. lett . * 95 * , 073102 ( 2009 ) .
j. bergli , y. m. galperin , and b. l. altshuler , new journal of physics * 11 * , 025002 ( 2009 ) .
e. paladino _ et al .
_ , phys .
rev .
lett . * 88 * , 228304 ( 2002 ) .
l. cywinski , w. m. witzel , and s. das sarma , phys .
rev .
b * 79 * , 245314 ( 2009 ) .
d. m. fleetwood , x. d. schrimpf , and s. t. pantelides , ieee trans .
nucl .
sci . * 49 * , 2674 ( 2002 ) ; g. i. wirth _ et al .
_ , ieee trans .
el .
dev . * 52 * , 1576 ( 2005 ) .
m. pioro - ladrire _ et al .
_ , phys .
rev .
b * 72 * , 115331 ( 2005 ) .
i. neder _ et al .
_ , arxiv:1103.4862 [ cond-mat.mes-hall ] ( unpublished ) .
guy ramon ( unpublished ) . |
the origin of high - energy cosmic - rays remains unclear in spite of the long history of cosmic ray study . while supernova remnants ( snrs ) are the favoured site for cosmic rays up to 10@xmath10ev , as they satisfy the required energy input rate to the galaxy and have a size comparable to the larmor radius of the highest energy particles , direct supporting evidence is sparse .
recently detections of gev gamma - rays from the regions near several snrs have been reported by egret ( esposito et al .
those results might be evidences of the assumed snr origin of cosmic rays .
however recent observations by the whipple group of six snrs , including three of egret sources , have given upper limits at hundreds of gev energy region ( buckley et al .
1998 ) which are below the expected flux from shock acceleration theory ( drury , aharonian , & vlk 1994 ; naito & takahara 1994 ) , unless cutoffs in the particle spectrum occur ( sturner et al . 1997 ; gaisser , protheroe , and stanev 1997 ) . on the other hand ,
intense non - thermal x - ray emission from the rims of type ia snr sn1006 ( g327.6 + 14.6 ) has been observed by asca ( koyama et al .
1995 ) , and rosat ( willingale et al . 1996 ) , which is considered , by attributing the emission to synchrotron radiation , to be strong evidence of an existence of high energy electrons up to @xmath11100 tev .
sn1006 is a typical shell - type snr which has no apparent central engine for high energy particles such as a neutron star or a black hole .
nevertheless , the existence of very high energy particles in a type ia snr is widely accepted from shock acceleration theory ( blandford & eichler 1987 ; jones & ellison 1991 ) .
if so , tev gamma rays would also be expected from inverse compton scattering ( ic ) of low energy photons ( mostly attributable to the 2.7 k cosmic background photons ) by these electrons . by assuming a value for the magnetic field strength ( @xmath12 ) in the emission region of the snr , several theorists ( pohl 1996 ; mastichiadis 1996 ; mastichiadis & de jager 1996 ; yoshida & yanagita 1997 ) calculated the expected spectra of tev gamma rays using the observed radio / x - ray spectra .
an observation of tev gamma rays would thus provide not only further direct evidence of the existence of very high energy electrons but also other important information such as the strength of the magnetic field and the diffusion coefficient of the shock acceleration . with this motivation ,
sn1006 was observed by the cangaroo imaging air erenkov telescope in march and june 1996 , and march and april 1997 .
the observations were made with the 3.8 m diameter erenkov imaging telescope of the cangaroo collaboration ( patterson & kifune 1992 ; hara et al .
1993 ) near woomera , south australia ( 136@xmath1347 e and31@xmath1306 s ) .
the 3.8 m alt - azimuth mounted telescope had a @xmath113 tev threshold for detecting gamma rays near 70@xmath13 elevation in the 1996 observations .
the 3.8 m mirror was recoated in 1996 october , and its reflectivity improved from 45% to more than 80% , decreasing the threshold energy by about a factor of two . a multi - pixel camera consisting of 256 square photomultiplier tubes , arranged in an array of [email protected] steps ,
has a total field of view ( fov ) of about 3@xmath13 ( hara et al .
1993 ) .
sn1006 was observed for 28 hours ( on - source ) and 18 hours ( off - source ) in april and june 1996 .
both of the north east ( ne ) and the south west ( sw ) rims were located within the fov . by monitoring the single counting rate in each phototube
, we were able to track the passage of the star ( magnitude 3.13 ) within the fov , and the pointing of the telescope was monitored to an accuracy of [email protected] using the trajectory of this star . in march and april 1997 , in an effort to confirm the 1996 result
, we made additional observations of 34 hours ( on source ) and 29 hours ( off - source ) with the same tracking as in june 1996 .
the imaging analysis of the data is based on the usual parameterization of the elongated shape of the erenkov light image as : `` width'',``length '' , `` distance '' ( location ) , `` conc '' ( shape ) , and the image orientation angle @xmath14 ( hillas 1985 ; weekes et al .
1989 ) . in the @xmath14 distribution of the events selected by the imaging analysis ,
the peak appearing around the origin @xmath15 in the on - source data is attributed to @xmath16-rays from the target position , and the number of background events under the peak was estimated from the flat region of the @xmath14 distribution ( 3090 ) in the on - source data . here
off - source data were used to verify the non - existence of any peculiar structure in the @xmath14 plot around the origin not due to gamma - ray events .
the application of this technique to data recorded with the cangaroo telescope has , to date , resulted in the detection of tev gamma rays from psr1706 - 44 ( kifune et al .
1995 ) and the nebula surrounding the crab ( tanimori et al . 1994
; tanimori et al .
1998 ) . from the results on the previously observed objects ,
the position of a gamma - ray point source can be determined to an accuracy of 0.1@xmath13 .
the point spread function ( psf ) of the cangaroo telescope is estimated to have a standard deviation of [email protected] when fitted with a gaussian function .
the hard x - ray profile of the ne rim observed by asca suggests that the tev gamma rays may emanate from an extended area over a few times the psf of the cangaroo telescope ( several tenths of a degree in extent ) . in order to search the emission region of tev gamma rays in sn1006 ,
significances of peaked events with @xmath17 were calculated at all grid - points in [email protected] steps in the fov which is half of the standard deviation of the psf . the source point in the ne rim giving the most significant @xmath14 peak ( @xmath18 ) was found at the maximum flux point in the 210kev band of the asca data .
the @xmath14 plot of the selected gamma - ray - like events at the x - ray maximum flux point is shown in fig.[fig : f1]a .
clear peaks due to an excess of gamma - ray events are seen at @xmath19 0@xmath13 for the on - source data of april and june but not for the off - source data . at this x - ray flux maximum point , the statistical significances of these peaks are estimated to be 3.0@xmath20 in april , 4.7@xmath20 in june , and 5.3@xmath20 in total , using the definition mentioned above .
the resulting contour map of significances is shown in fig.[fig : f2]a , in which the contours of the hard x - ray flux and the maximum flux point in the 210kev band of the asca data also are overlaid as solid bold - lines and marked by a cross , respectively .
the region showing significant tev gamma - ray emission extends along the ridge of the ne rim over the psf of the telescope , and matches the x - ray image fairly well . in march and april 1997 , in order to confirm the 1996 result , we made additional observations .
figure [ fig : f1]b shows the @xmath14 distribution of gamma - ray like events selected by the same procedure as used for the 1996 data .
a clear peak ( @xmath2115@xmath13 ) was observed again with the significance of 7.7@xmath20 at the maximum hard x - ray flux point of asca .
the improvement of the detection significance was due to the twice increase of the reflectivity of the mirror .
thus the tev gamma - ray emission from the ne rim of sn1006 has been confirmed ( tanimori et al .
figure [ fig : f2]b shows the contour map of significances .
the tev gamma - ray emission region also looks elongated along the ridge of the ne rim , although the profile of 1997 data is not as extend as that from the 1996 data . in order to verify whether the emission region is extended or not , the profiles of the ne rim in 1996 and 1997 data were fit using a superposition of two psfs located along the ridge of the ne rim and also using a single psf .
although the 1996 result favours the fit using a superposition of two psfs against a single psf , the 1997 data , with better statistics , does not show significant improvement for a superposition of two psfs . from these significance maps
we therefore can not claim the extent of the tev gamma - ray emission region quantitatively .
further study is required .
the threshold energy for the observed gamma rays was determined from monte carlo simulations , as the maximum of the product of the differential flux times the effective collecting area ; the latter is a function of gamma - ray energy .
compared the 1997 result to the 1996 result , the threshold energy was decreased about a factor of two , and the number of the detected gamma - like events was increased twice .
this indicates a differential photon spectral index @xmath22 of 2 around a few tev ( @xmath23 in photons c@xmath24s@xmath2tev@xmath2 ) .
also the @xmath14 peak due to gamma - ray like events can be seen even in about ten times higher energy region than the threshold energy in the 1997 data , which means that there exists little effect of the spectral cut - off in the observed energy region .
therefore the differential spectral index was simply assumed to be @xmath25 in calculating the integral fluxes of our data .
results of the monte carlo simulation give a threshold gamma - ray energy of 3@xmath26 tev and an effective area of approximately 6.6@xmath27 10@xmath28 @xmath29 for the 1996 observation of sn1006 . approximating the emission as coming from a single point source at the maximum flux point in the ne rim , the integral gamma - ray flux for the 1996 observations was calculated to be @xmath30 @xmath1 s@xmath2 ( @xmath3 tev ) .
the threshold energy for the 1997 observations was also estimated to be @xmath31 tev . using similar approximations to those above
, the gamma - ray integral flux for the 1997 observation was calculated to be @xmath4 @xmath1 s@xmath2 ( @xmath5 tev ) .
in those fluxes the first and second errors are statistical and systematic respectively .
the systematic error mainly arises from the uncertainty of the absolute threshold energy .
a larger flux would be obtained if the emission extends wider than the psf of our detection .
also the systematic error due to the assumed differential spectrum was evaluated : considering the uncertainties in shock acceleration models such as nonlinear effects and energy cut off in the electron spectrum ( jones & ellison 1991 ) , we varied the spectral index from 1.2 to 4.0 . in this range , the integral flux changes by about 2% ( 1.2 ) to + 20% ( 4.0 ) , which is relatively smaller than that due to the uncertainty of the absolute threshold energy .
no significant excess is evident in fig .
[ fig : f2]a near the position of the maximum x - ray flux from the sw rim .
the x - ray observation by asca indicates that the integral flux of hard x - rays ( @xmath32 2 kev ) in the ne rim occupies @xmath1160% of the whole hard x - ray flux emitted from sn1006 ( ozaki 1997 ) .
the current analysis method of using the @xmath14 distribution has difficulty in separating two emission regions as close as [email protected] of the ne and sw rims .
the weaker emission might be hindered by the stronger one from the ne rim .
thus , we set an upper limit on the tev gamma - ray emission from the sw rim , estimated to be @xmath6 @xmath1 s@xmath2 ( @xmath5 tev , 95% cl ) from the @xmath14 distribution in the 1997 data at the position of the maximum asca flux in the sw rim .
this detection of tev gamma rays from sn1006 presents a convincing confirmation of the shock acceleration mechanism for very high energy particles up to @xmath11100 tev in a snr .
the tev gamma ray emission region is observed to likely extend over @xmath1130 arcmin along the ridge of the ne rim . from the non - thermal x - ray observation ,
the detected tev gamma rays are readily presumed to be generated by ic scattering of very high energy electrons on @xmath33k cosmic background photons .
all of the calculated fluxes of tev gamma rays based on those assumptions are consistent with the tev gamma - ray fluxes obtained by assuming that the magnetic field strength ( @xmath12 ) in the emission region of the snr is around 10@xmath9 g .
one model ( yoshida & yanagita 1997 ) calculates the expected tev gamma - ray spectrum as a function of the strength of the magnetic field assuming a power law with exponential cutoff energy spectrum for electrons where the value of necessary parameters are determined by fitting the observed radio and x - ray synchrotron emissions .
the observed fluxes of tev gamma - rays ( @xmath34 tev and @xmath35 tev ) fit well if we take @xmath36 g in the model as shown in fig.[fig : f3 ] .
the other candidate of the production mechanism of tev gamma - ray emission is a decay of neutral pions induced by high energy protons accelerated in the snr .
however , we can neglect the flux from the @xmath37 decay due to following arguments . since sn1006 ( g327.6 + 14.6 )
is located above the galactic plane , the matter density at the shock is low ( @xmath110.4 @xmath38 : willingale et al .
1996 ) so that the expected flux will be about a factor of ten less than the observed flux .
the upper limit for gev gamma - ray emission from the egret archive data is also consistent with the ic model .
thus , the detected gamma - rays are likely to be explained by ic radiation from electrons , and our result testifies to the existence of the very high energy electrons of more than several times 10 tev in sn1006 .
the highest energy of non - thermal electrons can be estimated from the turning point in the synchrotron spectrum and the resultant magnetic fields .
although the turning energy of sn1006 is not yet precisely determined in recent observations , a 1 kev photon from synchrotron radiation corresponds to an electron energy of @xmath39 tev for @xmath40 g .
these values of highest energy and field strength , and 1000 years of life time for sn1006 are almost consistent with shock acceleration theory .
observations of some other bands and evolutional theories of the snr are required to confirm the highest accelerated energy more precisely ( sturner et al .
1997 ; gaisser , protheroe , and stanev 1997 ) .
the observed concentration of hard x - rays and gamma - rays emissions into the rims in sn1006 also suggests the possibility that the relation between the direction of the magnetic field and the shock front may determine the efficiency of particle acceleration .
reynolds ( reynolds 1996 ) pointed out that the magnetic field in the upstream of the shock of the rims in sn1006 is likely to be parallel to the shock front , which may show the predominance of highly oblique shocks , where the efficiency of shock acceleration is improved ( jokipii 1987 ; naito & takahara 1995 ) .
searches for tev gamma - ray emission from six snrs in the northern hemisphere have been carried out using a large imaging air erenkov telescope by the whipple group ( buckley et al .
1998 ) , and turned out unsuccessfully .
sn1006 is so far the only snr in which the existence of very high energy electrons up to @xmath11 100 tev is suspected from x - ray data .
recently non - thermal hard x - ray emissions from several snrs have been observed ( koyama et al .
1997 ; keohane et al .
1997 ; allen et al .
1997 ) , implying the existence of electrons up to a few tens of tev .
it is clear that more efforts for detecting tev gamma rays from snrs are necessary to understand the shock acceleration mechanism in more detail and therefore the origin of cosmic rays .
we thank prof .
k. koyama , prof .
j. nishimura , dr . m. ozaki , and dr .
o. de jager for their fruitful discussions .
this work is supported by a grant - in - aid in scientific research of the japan ministry of education , science , sports and culture , the australian research council and international science and technology program and the sumitomo foundation .
the receipt of jsps research fellowships ( p.g.e .
, g.j.t . and
t. yoshikoshi ) is also acknowledged .
allen , g. e. , et al .
1997 , , 487 , l97 blandford , r.d . , &
eichler , d. 1987 , phys .
, 154 , 1 buckley , j.h .
, et al .
1998 , , 329 , 639 drury , l.oc . ,
aharonian , f.a . , &
vlk , h.j .
1994 287 , 959 esposito , j.a .
, et al . 1996 , , 461 , 820 gaisser , t.k . , protheroe , r.j . ,
& stanev , t. 1998 , , 492 , 219 hara , t. , et al .
1993 , nucl .
instr . and meth . ,
a332 , 300 jokipii , j.r . , 1987 , , 313 , 842 jones , f.c . , & ellison , d.c .
1991 , space sci .
rev . , 58 , 259 keohane , j. w. , et al .
1997 , , 484 , 350 kifune , t. , et al .
1995 , , 438 , l91 koyama , k. , et al .
1995 , , 378 , 255 koyama , k. , et al .
1997 , , 49 , l7 mastichiadis , a. , & de jager , o.c .
1996 , , 311 , l5 naito , t. , & takahara , f. 1994 , j. phys .
g , 20 , 477 naito , t. , & takahara , f. 1995 , , 275 , 1077 ozaki , m. 1997 , private communication patterson , j. r. , & kifune , t. 1992 , australian and new zealand physicist , 29 , 58 pohl , m. 1996 , , 307 , 57 reynolds , s.p .
1996 , , 459 , l13 sturner , s. j. , et al .
1997 , , 490 , 619 tanimori , t. , et al .
1998 , , 492 , l33 tanimori , t. , et al .
1997 , iau circ .
6706 weekes , t.c .
, et al . 1989 , , 342 , 379 willingale , r. , et al .
1996 , , 278 , 749 yoshida , t. , & yanagita , s. 1997 , in proc .
2nd integral workshop ` transparent universe ' , esa sp-382 , 85 yoshikoshi , t. , et al .
1997 , , 487 , l65 | this paper reports the first discovery of tev gamma - ray emission from a supernova remnant made with the cangaroo 3.8 m telescope .
tev gamma rays were detected at the sky position and extension coincident with the north - east ( ne ) rim of shell - type supernova remnant ( snr ) sn1006 ( type ia ) .
sn1006 has been a most likely candidate for an extended tev gamma - ray source , since the clear synchrotron x - ray emission from the rims was recently observed by asca ( koyama et al .
1995 ) , which is a strong evidence of the existence of very high energy electrons up to hundreds of tev in the snr .
the observed tev gamma - ray flux was @xmath0 @xmath1 s@xmath2 ( @xmath3 tev ) and @xmath4 @xmath1 s@xmath2 ( @xmath5 tev ) from the 1996 and 1997 observations , respectively .
also we set an upper limit on the tev gamma - ray emission from the sw rim , estimated to be @xmath6 @xmath1 s@xmath2 ( @xmath5 tev , 95% cl ) in the 1997 data .
the tev gamma rays can be attributed to the 2.7 k cosmic background photons up - scattered by electrons of energies up to about 10@xmath7 ev by the inverse compton ( ic ) process .
the observed flux of the tev gamma rays , together with that of the non - thermal x - rays , gives firm constraints on the acceleration process in the snr shell ; a magnetic field of @xmath8 @xmath9 g is inferred from both the synchrotron x - rays and inverse compton tev gamma - rays , which gives entirely consistent mechanisms that electrons of energies up to 10@xmath7 ev are produced via the shock acceleration in sn1006 . |
the deconfinement / chiral symmetry restoration phase transition in ultrarelativistic heavy ion collisions is inherently a multiparticle phenomenon .
therefore , event - by - event analysis of multiparticle hadronic observables is of paramount interest .
we carry out a texture , or local fluctuation , analysis to determine the correlation / fluctuation content of single events , and analyze the scale composition of the correlations .
the idea to look at particle distributions in rapidity @xmath1 to search for critical behaviour was proposed @xcite based upon a ginzburg - landau type of multihadron production theory @xcite , where a random hadronic field @xmath2 plays the role of an order parameter in a hadronization transition . enhanced large scale correlations of hadrons in @xmath1 at
the phase transition would signal critical fluctuations in the order parameter .
stephanov and coworkers @xcite indicated a second order qcd phase transition point which should exist under certain initial conditions , within the reach of today s experiments . in our work
, a power spectrum analysis of event texture in pseudorapidity @xmath3 and azimuthal angle @xmath4 , based on a discrete wavelet transformation ( dwt)@xcite , is performed on a number of large event ensembles sampled according to their multiplicity , thereby studying the impact parameter dependence of the observables .
dwt quantifies contributions of different @xmath4 and @xmath3 scales into the event s overall texture , thus testing the possible large scale enhancement .
= 7.5 cm = 7.5 cm
ionization energy loss of charged particles was measured in 512 silicon pads , with radial granularity of 16 and azimuthal granularity of 32 .
the pads were read out by amplex @xcite chips , one chip per sector .
@xmath5-electrons , produced by the @xmath6 beam traversing the target , were swept away to one side by a dipole magnetic field ( @xmath7 tl ) . only the @xmath5-electron - free side was used in the analysis .
empty target runs were used to measure the background .
cross - talk in the detector was evaluated off - line .
track density in an individual event @xmath8 is represented by a 2d array of the calibrated digitized amplitudes of the channels ( an _ amplitude array _ ) .
this is expanded into a basis of _ _ haar wavelet__@xcite functions @xmath9 , orthogonal with respect to scale fineness @xmath10 and location @xmath11 , with the @xmath12 index numbering three modes of texture direction sensitivity @xmath4 , @xmath3 , and @xmath13 for azimuthal , pseudorapidity and diagonal .
is constructed from 1d functions in the following way : @xmath14 , where for any variable @xmath15 , the wavelet function @xmath16 , and the scaling function @xmath17 for @xmath18 and 0 otherwise . ] with our binned detector , the act of data taking is the first stage of the haar wavelet transformation ( not true for any other wavelet ) . from the squared expansion coefficients a _ power spectrum _ is formed : @xmath19 we used waili @xcite software to obtain the wavelet expansions .
more experimental and algebraic details can be found in @xcite and the references therein .
figure [ compare ] shows that the power spectra in central pb+pb collisions are indeed enhanced on the coarse scale .
it is necessary to eliminate `` trivial '' and detector related contributions to this enhancement .
event mixing is done by taking different channels from different events . in order to reproduce the electronics cross - talk effects in the mixed event sample
, mixing is done sector - wise , i.e. the sectors constitute the subevents subjected to the event number scrambling .
the mixed events preserve the texture associated with detector position offsets , the inherent @xmath20 shape , cross - talk and dead channels . this is _
static _ texture as it reproduces its pattern event after event ; we are interested in _ dynamic _ texture .
we reduce sources of static texture in the power spectra by empty target subtraction and by subtraction of the mixed events power spectra , thus obtaining the _ texture correlation _ @xmath21 .
this quantity , normalized to the @xmath22 fluctuation of @xmath23 , is used to characterize the relative strength of local fluctuations in an event .
its distribution for different @xmath12 is plotted on figure [ limit ] in an integral way , i.e. as an @xmath24 graph where for every @xmath15 , @xmath25 is the fraction of the distribution above @xmath15 .
the _ confidence level _ with which local fluctuations of a strength @xmath15 can be excluded is then @xmath26 .
fluctuations greater than @xmath27 are excluded in the azimuthal and pseudorapidity modes with 90% and 95% confidence , respectively .
the monotonic fall of the curve is consistent with absence of abnormal subsamples in the data .
figure [ multi_dep ] summarizes texture correlation data for different multiplicity events .
the systematic errors were evaluated by removing the @xmath6 target and switching magnetic field polarity to expose the analyzed side of the detector to @xmath5-electrons , while minimizing nuclear interactions . correlations ( i.e. deviations of @xmath28 from @xmath23 ) in such events measure the remaining systematic uncertainties .
thus , this component of the systematic error is signed , and the systematic errors are asymmetric
. the other component ( significant only on the coarsest scale ) is the uncertainty of our knowledge of the beam s geometrical cross - section .
a monte carlo simulation was performed , including known background texture effects and the subtraction of mixed simulated events . the finite beam size effect , irreducible by the subtraction methods ,
is thereby taken into account .
the finite beam size effect explains the rise of rqmd points with @xmath20 in fig .
[ multi_dep ] .
finally , the sensitivity of the method is evaluated ( see figure [ sensitivity ] ) in mc by generating multiple fireball events with given mean number of particles ( charged and neutral ) , @xmath29 , per fireball .
the total @xmath30 of each fireball is 0 ; its total @xmath31 is chosen to simulate longitudinal flow by lorentz - boosting the fireballs along the @xmath32 direction , keeping the total @xmath33 of an event at 0 . by varying @xmath29 per fireball
, one varies `` grain coarseness '' of the event texture in @xmath3 .
the data are consistent with clustering of @xmath29 per fireball below 50 .
this novel method of event - by - event analysis , applied to the sps @xmath34 data , does not reveal any evidence of critical phenomena .
the authors thank n.antoniou , i.dremin , e.shuryak , m.stephanov , and t.trainor for illuminating discussions .
this work was supported by the austrian fonds zur frderung der wissenschaftlichen forschung ; the science research council of denmark ; the japanese society for the promotion of science ; the ministry of education , science and culture , japan ; the science research council of sweden ; the us department of energy and the national science foundation .
99 d.j.scalapino , r.l.sugar , phys . rev .
* d8 * , no 7 , 2284 - 2294 ( 1973 ) p.carruthers , i.sarcevic , phys .
* b189 * , no 4 , 442 - 448 ( 1987 ) m.stephanov , k.rajagopal , e.shuryak , phys . rev .
* 81 * 4816 ( 1998 ) i.daubechies , ten lectures on wavelets , siam , 1992 e. beuville _
et al . _ , nim * a288 * , 157 - 167 ( 1990 ) j.pando , l-z.fang , phys .
rev . * e57 * , 3593 - 3601 ( 1998 ) ; l-z.fang , astro - ph/9701228 .
g. uytterhoeven _
et al . _ , waili : wavelets with integer lifting .
tw report 262 , department of computer science , katholieke universiteit leuven , belgium , july 1997 . | na44 uses a 512 channel si pad array covering @xmath0 to study charged hadron production in 158 a gev pb+pb collisions at the cern sps .
we apply a multiresolution analysis , based on a discrete wavelet transformation , to probe the texture of particle distributions event - by - event , by simultaneous localization of features in space and scale . scanning a broad range of multiplicities ,
we look for a possible critical behaviour in the power spectra of local density fluctuations .
the data are compared with detailed simulations of detector response , using heavy ion event generators , and with a reference sample created via event mixing .
an upper limit is set on the probability and magnitude of dynamical fluctuations . |
entanglement is a central concept in quantum theory and an important resource in quantum information theory .
different entanglement measures have been introduced to understand the structure of entanglement between finite - dimensional systems .
one such measure is entanglement entropy , which can also be defined for pure states of extended quantum systems .
since all measurements on extended systems are performed within some finite region of space , it is natural in the relativistic context to study entanglement between localized subsystems , i.e. subsystems associated with spacelike separated regions of spacetime . although natural , this is also problematic , because relativistic field theories typically have infinite entanglement entropy between a region of space and its complement at fixed time @xcite .
indeed , the separation of the degrees of freedom of a quantum field into a finite region of space @xmath0 and its complement @xmath1 implies a specific mapping associating @xmath0 and @xmath1 with commuting subalgebras of the algebra of observables , called a localization scheme @xcite .
different localization schemes correspond to different tensor product structures ( tps ) of the total hilbert space .
since entanglement is a property defined between subsystems , different localization schemes yield different values for entanglement between @xmath0 and @xmath1 @xcite . a standard localization scheme associates to space regions the field operators and their conjugates therein defined .
the vacuum state is entangled with respect to the tps induced by the standard localization scheme , hence any two localized subsystems are correlated by modes at their boundary .
consequently , the entropy of entanglement between @xmath0 and @xmath1 diverges due to the contribution of ultraviolet ( uv ) modes .
this peculiar behavior of entanglement is deeply rooted in the algebraic structure of relativistic quantum field theory ( qft ) .
indeed , it is a direct consequence of the reeh - schlieder theorem @xcite that all finite - energy states maximally violate bell - like inequalities @xcite .
an operational content is given to this violation in @xcite : whenever the local algebras of observables of a bipartite system are not type i von neumann algebras , maximally entangled states have infinite one - copy entanglement . for systems with type
i local algebras of observables , states with infinite entropy of entanglement are trace - norm dense in state space @xcite .
it follows from these arguments that entanglement measures for bipartite states with localized subsystems typically diverge at all energy scales when these are analyzed at a more fundamental level using qft .
such subsystems do not yet correspond to the ones defined in the nonrelativistic regime , _
e.g. _ qubits or trapped ions in quantum information protocols .
the latter are described by finite dimensional hilbert spaces and they generally produce finite results for entanglement measures .
one expects that in the low - energy limit the degrees of freedom describing a region of space should simply correspond to the degrees of freedom of the particles therein contained .
however , as a consequence of the reeh - schlieder theorem one can not define local number operators , therefore finite - energy states can not be localized @xcite , making such a correspondence impossible .
hence a conceptual tension between the qft description of entanglement for low - energy experiments and a description using nonrelativistic quantum theory . in this paper
, we introduce a novel approach to reconcile these two disagreeing notions of entanglement .
consider a scenario in which all possible field measurements are limited by some minimal spatial resolution @xmath2 , thus restricting the algebra of observables to coarse - grained fields .
the coarse - graining parameter @xmath2 has a clear operational meaning : since in practical situations it is impossible to resolve points in space with arbitrary precision , any realistic measurement of a field necessarily consists of a sample of a finite number of points , where each point corresponds to a finite region of space .
we show that in the limit @xmath3 , @xmath4 being the mass of a klein - gordon field , states with a finite number of particles become localized , allowing an identification between a region of space and the nonrelativistic degrees of freedom of the particles therein contained , and that the renormalized entropy of finite - energy states reduces to the one calculated in nonrelativistic quantum mechanics .
this provides the missing controlled transition from the qft picture of entanglement to entanglement in nonrelativistic quantum theory .
consider at fixed time a finite region of space @xmath0 and its complement @xmath1 .
region @xmath0 has two complementary descriptions : classical general relativity identifies it with a submanifold of minkowski spacetime , but as a quantum subsystem , @xmath0 is described by a hilbert space @xmath5 , which is a factor in the tensor product decomposition @xmath6 of the total hilbert space of the field theory under investigation .
suppose that the field is in a state @xmath7 .
the results of measurements to be performed in region @xmath0 are described by the reduced density matrix obtained by tracing out the degrees of freedom outside @xmath0 : @xmath8 .
the von neumann entropy associated with region @xmath0 is then defined as @xmath9 .
this quantity typically requires some uv regulator in order to be well defined .
thus , in @xcite , a uv cutoff @xmath10 is introduced at the boundary between @xmath0 and @xmath1 , and in @xcite a qft is defined on a lattice of spacing @xmath10 .
references @xcite show that for the vacuum state , the entropy of entanglement associated with region @xmath0 can be written as : @xmath11 where @xmath12 is the area of the boundary between @xmath0 and @xmath1 , @xmath4 the mass of the field , @xmath13 an infrared cutoff and @xmath14 some slowly varying function . for finite - energy states , power - law correction terms need to be added @xcite .
these expressions diverge in the continuum limit for @xmath15 and , more generally , no cutoff - independent low - energy limit of the entropy can be derived using these approaches .
a renormalization technique proposed in @xcite leads to states of negative entropy , which is not expected for a physically meaningful concept of state entropy .
yet another approach consists in introducing a physical model of the measurement apparatus .
one then derives an effective low - energy model of the measurement apparatus insensitive to vacuum entanglement of the underlying qft @xcite .
however , this approach is strongly dependent on the choice of the model and does not provide a general and clear transition from the qft picture of entanglement to entanglement in nonrelativistic quantum theory .
for simplicity , we consider a neutral klein - gordon field of mass @xmath4 in one space dimension at fixed time ( we put @xmath16 ) .
the algebra of local observables for the klein - gordon field is generated by the canonical field operators : @xmath17 where @xmath18 and @xmath19 creates a field excitation of momentum @xmath20 .
the vacuum is defined by : @xmath21 assume that the resolution for distinguishing different points in space is bounded by some minimal length @xmath2 .
the algebra of observables that are accessible under such conditions is generated by the coarse - grained field operators ( see fig.[schema ] ) : @xmath22 function @xmath23 describes the detection profile : @xmath24this choice of profile is natural if we interpret coarse graining as arising from a random error in the identification of a point in space . more generally , for any profile with a typical length @xmath25 , consider intervals of length @xmath2 on which the profile is approximately constant .
one can then convolute such a profile with a gaussian of variance @xmath2 and consider the limit @xmath26 instead of @xmath27 . .
this source of error is implemented by restricting the observable degrees of freedom to those accessible via measurement of coarse - grained operators .
neighboring profiles define different subsystems only if their separation @xmath28 verifies @xmath29 . under this condition ,
entanglement between neigboring profiles is a well - defined notion .
we show that for finite - energy states , this entanglement reduces to the one calculated in nonrelativistic quantum theory . ]
define the operators : @xmath30where @xmath28 is the distance between neighbouring profiles .
if @xmath31 , they verify canonical commutation relations : @xmath32\sim i\delta_{j\,k}.\]]imposing is equivalent to saying that the operators @xmath33 generate commuting subalgebras .
two commuting subalgebras of observables @xmath34 and @xmath35 that generate the whole algebra of observables induce a tps on the hilbert space of states : @xmath36 such that @xmath37 , @xmath38 @xcite .
the operators generate only a strict subalgebra of the entire algebra of field observables , because under coarse graining some possible observables are inaccessible .
the whole algebra can be recovered by completing the set of functions @xmath39 up to an orthonormal basis in @xmath40 which , convoluted with the field operators , defines a linear canonical transformation of modes .
thus , the algebra generated by the coarse - grained observables defines a decomposition of the total hilbert space @xmath41 , where @xmath42 are the coarse - grained , hence accessible , and @xmath43 the fine - grained inaccessible degrees of freedom .
the restriction to coarse - grained observables is therefore equivalent to tracing out subsystem @xmath43 , and operators @xmath44 define distinct subsystems on @xmath45 , each of which is isomorphic to a one - dimensional harmonic oscillator .
thus , we can define on @xmath45 the coarse - grained ladder operators : @xmath46which verify @xmath47 \sim \delta_{jk}$ ] .
parameter @xmath48 has the dimension of mass .
for a massive klein - gordon field , it is natural to take @xmath49 .
indeed , one can alternatively generate the local observables algebra with the ladder operators : @xmath50their coarse - grained versions correspond to operators in with @xmath49 .
we recall that the newton - wigner ( nw ) annihilation and creation operators are respectively defined as the fourier transforms of the momentum annihilation and creation operators @xcite : @xmath51the nw operators define a localization scheme @xcite and are expressed in terms of the local fields as follows : @xmath52 , \\
\hat a_{nw}^{\dag}(x ) = & \frac{1}{\sqrt{2}}\int dy\left[r(x - y)\hat \phi(y ) -i r^{-1}(x - y)\hat \pi(y)\right ] , \end{aligned}\ ] ] where we have introduced the functions : @xmath53operators @xmath54 annihilate the global vacuum , therefore the global vacuum is a product state of local vacua .
more generally , if local degrees of freedom are associated with the nw operators instead of the standard local fields , entropy of entanglement of the vacuum state is zero and entropy of finite - energy states , such as thermal states , becomes finite @xcite .
however , identifying local degrees of freedom with nw operators at a fundamental level is problematic : the hamiltonian of the field , expressed in terms of nw operators , is non - local .
we do not address here the question of which localization scheme should be chosen at a fundamental level @xcite .
instead we show that , under coarse graining , the entanglement properties of the nw fields for finite - energy states effectively hold , irrespective of the fundamental choice of local observables .
we now compare the algebra of coarse - grained observables generated by with the algebra generated by the following coarse - grained nw operators : @xmath55 computations show that : @xmath56 , \end{split}\ ] ] where : @xmath57 , \\ r_{\epsilon}(x)= & \int dy g_{\epsilon}(x - y)r(y ) ,
\\ r^{-1}_{\epsilon}(x)= & \int dy g_{\epsilon}(x - y)r^{-1}(y ) .
\end{split}\]]in the limit of poor space resolution , the coarse - grained nw operators become indistinguishable from the coarse - grained local ladder operators since : @xmath58 , \end{aligned}\]]and in the limit where the minimal resolvable distances are much larger than the compton wavelength , @xmath59 , the gaussian @xmath60 verifies @xmath61 for @xmath62 and @xmath63 otherwise .
thus , in we have to integrate only over small values of @xmath20 .
we find : @xmath64 this result , plugged back into , gives : @xmath65 in the limit @xmath3 , the coarse - grained nw operators still annihilate the global vacuum , hence the latter is a product state of effective local vacua .
equation then shows that the global vacuum is also a product state for the coarse - grained field operators .
this implies that , in the limit of poor spatial resolution of detectors , an excitation localized around point @xmath66 " is effectively described by applying the creation operator @xmath67 to the global vacuum @xmath68 .
therefore , any one - particle state @xmath69 can be effectively described as a sum @xmath70 , where @xmath71 is a function verifying @xmath72 and @xmath73 its fourier transform . as a consequence , such a state ( which can not be interpreted as localized in qft unless it has infinite energy )
can now be properly interpreted as localized , allowing a mapping between the description of a region of space in qft and an effective description that only includes the nonrelativistic degrees of freedom therein contained .
hence , the structure of entanglement of any state with a finite number of excitations reduces to the entanglement between localized particles , i.e. to the standard , nonrelativistic , picture of entanglement . in particular , the entropy of entanglement of such states is upper bounded by the number of excitations times a factor describing how many states are available to each excitation ( see the appendix ) . since finite - energy states correspond to states with a finite number of excitations , this result provides a controlled transition from the qft picture of entanglement of finite - energy states to the nonrelativistic quantum theory one . as an example , consider two mesons or two atoms with integer spin in a singlet state localized around points @xmath74 and @xmath66 " . in the qft picture ,
entanglement between the region around point @xmath74 " containing one meson with the rest of the system is infinite . under the constraint of a bounded spatial resolution of detectors ,
the effective description of such a system in qft is : @xmath75|\omega\rangle \\ = & \frac{1}{\sqrt{2}}\left(|0\rangle_1\cdots|0\rangle_{i-1}|\uparrow \rangle_i|0\rangle_{i+1}\cdots|0\rangle_{j-1 } |\downarrow\rangle_j |0\rangle_{j+1}\cdots\right.\\ & \left.+\right|0\rangle_1\cdots|0\rangle_{i-1}|\downarrow \rangle_i|0\rangle_{i+1}\cdots|0\rangle_{j-1 } |\uparrow\rangle_j |0\rangle_{j+1}\cdots ) , \end{aligned}\]]which is formally equivalent to the state : @xmath76.\]]the entropy of entanglement between the region around @xmath74 " and
the rest of the system is then @xmath77 , which is the expected value when modeling this system in nonrelativistic quantum theory .
note that by symmetry @xmath78 .
we have shown that in the limit of poor spatial resolution of the detectors , the entropy of entanglement of finite - energy states of a massive klein - gordon field reduces to the one calculated in nonrelativistic quantum theory .
the derivation was independent of any effective low - energy model for the detectors .
the results of this paper can be generalized to all noncritical bosonic systems , i.e. systems endowed with a finite length scale such as lattice models or models with local interactions and an energy gap ( a natural length scale is then provided by the lattice spacing and the correlation length respectively ) @xcite .
for critical systems , the correlation length diverges , hence different arguments are needed . for fermionic systems ,
an ambiguity in the definition of entanglement measures between subsystems arises due to the anticommutation of the creation and annihilation operators @xcite .
one can reformulate the problem of entanglement between localized subsystems in a purely algebraic way @xcite , and a possible extension of the coarse - graining procedure to the algebraic framework is under investigation .
we thank aslav brukner , magdalena zych and federico piazza for helpful discussions .
this work was supported by the european commission project q - essence ( no.248095 ) , the european commission project raquel , the john templeton foundation , fqxi , and the austrian science fund ( fwf ) through coqus , sfb foqus , and the individual project no.2462 .
consider at fixed time a finite region of space @xmath0 and its complement @xmath1 .
suppose that the field is in a state with @xmath79 excitations .
@xmath0 is decomposed into @xmath80 distinct regions @xmath81 , whose points are assumed to be nonresolvable because of the limited spatial resolution of the detectors .
an upper bound on the entropy of entanglement between subsystems @xmath0 and @xmath1 is given by the dimension of the subspace of an @xmath80-mode system containing any number of particles between 0 and @xmath79 : @xmath82 where : @xmath83is the dimension of the subspace with exactly @xmath84 particles .
this provides an upper bound on the entropy of entanglement between @xmath0 and @xmath1 for the @xmath79-particle state : @xmath85 if @xmath86 , @xmath87 , expressing the fact that the entropy of entanglement of such states is upper bounded by the number of excitations times a factor describing how many states are available to each excitation .
one can encode degrees of freedom other than position by changing the value of @xmath80 .
for example , if two polarization states are available to each excitation , one must double the value of @xmath80 .
37ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty http://journals.aps.org/prd/abstract/10.1103/physrevd.34.373 [ * * , ( ) ] http://journals.aps.org/prl/abstract/10.1103/physrevlett.71.666 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/0370269394910073 [ * * , ( ) ] http://iopscience.iop.org/1742-5468/2004/06/p06002/ [ ( ) ] http://journals.aps.org/rmp/abstract/10.1103/revmodphys.82.277 [ * * , ( ) ] http://www.jstor.org/discover/10.2307/188690?uid=3738016&uid=2&uid=4&sid=21104240546801 [ * * , ( ) ] http://www.jstor.org/discover/10.2307/3081027?uid=3738016&uid=2&uid=4&sid=21104240546801 [ * * , ( ) ] http://scitation.aip.org/content/aip/proceeding/aipcp/10.1063/1.2218238 [ , ( ) ] @noop _ _ , qg - ph ( ) http://arxiv.org/abs/0711.3048[[arxiv:0711.3048 ] ] http://link.springer.com/article/10.1007%2fs10701-009-9387-5 [ * * , ( ) ] http://journals.aps.org/prl/abstract/10.1103/physrevlett.87.077901 [ * * , ( ) ] http://journals.aps.org/prl/abstract/10.1103/physrevlett.92.060402 [ * * , ( ) ] http://link.springer.com/article/10.1007%2fbf02787889 [ * * , ( ) ] http://journals.aps.org/prd/abstract/10.1103/physrevd.10.3320 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/0375960185900933 [ * * , ( ) ] http://link.springer.com/article/10.1007%2fbf01207366 [ * * , ( ) ] http://link.springer.com/article/10.1007%2fbf02054660 [ * * , ( ) ] in @noop _ _ ( ) pp .
http://www.sciencedirect.com/science/article/pii/s1355219800000332 [ * * , ( ) ] in @noop _ _ ( ) pp . @noop * * , ( ) http://www.worldscientific.com/doi/abs/10.1142/s0129055x0600284x [ * * , ( ) ] http://iopscience.iop.org/0305-4470/35/17/307/ [ * * , ( ) ] http://journals.aps.org/prd/abstract/10.1103/physrevd.73.121701 [ * * , ( ) ] http://journals.aps.org/prd/abstract/10.1103/physrevd.77.064013 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/0550321394904022 [ * * , ( ) ] http://iopscience.iop.org/1367-2630/11/11/113006/ [ * * , ( ) ] http://journals.aps.org/prd/abstract/10.1103/physrevd.81.125019 [ * * , ( ) ] http://journals.aps.org/rmp/abstract/10.1103/revmodphys.21.400 [ ] http://journals.aps.org/prd/abstract/10.1103/physrevd.79.025006 [ * * , ( ) ] http://iopscience.iop.org/1742-5468/2007/08/p08024/ [ ( ) ] http://journals.aps.org/pra/abstract/10.1103/physreva.80.052104 [ * * , ( ) ] http://journals.aps.org/pra/abstract/10.1103/physreva.83.062323 [ * * , ( ) ] http://journals.aps.org/pra/abstract/10.1103/physreva.87.022338 [ * * , ( ) ] http://www.sciencedirect.com/science/article/pii/s0034487705800096 [ * * , ( ) ] http://scitation.aip.org/content/aip/journal/jmp/51/1/10.1063/1.3285046 [ * * , ( ) ] http://arxiv.org/abs/1401.2652 [ ] | entanglement is defined between subsystems of a quantum system , and at fixed time two regions of space can be viewed as two subsystems of a relativistic quantum field .
the entropy of entanglement between such subsystems is ill - defined unless an ultraviolet cutoff is introduced , but it still diverges in the continuum limit .
this behavior is generic for arbitrary finite - energy states , hence a conceptual tension with the finite entanglement entropy typical of nonrelativistic quantum systems .
we introduce a novel approach to explain the transition from infinite to finite entanglement , based on coarse graining the spatial resolution of the detectors measuring the field state .
we show that states with a finite number of particles become localized , allowing an identification between a region of space and the nonrelativistic degrees of freedom of the particles therein contained , and that the renormalized entropy of finite - energy states reduces to the entanglement entropy of nonrelativistic quantum mechanics . |
jovian radio emission demonstrates a variety of fine spectral and temporal structures .
the most studied are the fine structures in the decametric range ; their typical spectral patterns are described , e.g. , in the papers of @xcite , @xcite , @xcite , and references therein . the emission at lower frequencies ( i.e. , in the hectometric and kilometric ranges ) is less studied , because it can not be observed from the earth .
space - based observations have revealed that some fine spectral structures in these ranges look similar to the decametric phenomena , while a lot of low - frequency radio bursts with fine structure demonstrate different morphologies and probably have different formation mechanisms @xcite . to date
, the most detailed ( i.e. , with the highest temporal and spectral resolutions ) observations of the low - frequency jovian radio emission were performed by the _
cassini _ spacecraft in 2000/2001 , during its flyby of jupiter en route to saturn @xcite .
one of the most puzzling phenomena observed were the kilometric bursts ( with the frequency of @xmath0 khz ) that consisted of several parallel stripes of emission in the dynamic spectrum .
these bursts were rather infrequent and were recorded only a few times .
the most prominent example is shown in fig .
[ fig01 ] .
this dynamic spectrum was recorded by the wide - band receiver of the _ cassini _ radio and plasma wave science experiment , with the time and frequency resolutions of @xmath1 ms and @xmath2 hz , respectively @xcite . as can be seen in fig .
[ fig01 ] , the striped spectral structure is the most pronounced in the time interval from 03:00:00 to 03:00:40 scet , where one can see up to five emission stripes in the frequency range from 30 to 70 khz .
the stripes exhibit rapid irregular frequency changes while remaining nearly parallel to each other .
the stripes are not exactly equidistant : in general , the frequency intervals between the adjacent stripes slightly increase with the emission frequency @xcite ; both these intervals and the number of stripes vary with time . in the beginning of the considered event ( at 02:58:40
02:59:40 scet ) , one can see short quasi - periodic bursts with very fast frequency drift ( both the positive and negative drifts are observed ) superimposed on a weakly - pronounced striped structure ; the drift rate of the short bursts sometimes approaches infinity , so they look rather like broadband pulsations . the intensity of the burst shown in fig .
[ fig01 ] , after conversion into energy flux density @xcite and normalization to a distance of 1 au from the source , is up to @xmath3 w @xmath4 @xmath5 ; this is about two orders of magnitude higher than a typical ( rotation - averaged ) intensity of jovian radio emission at these frequencies @xcite .
the bursts with striped spectral structure seem to be a part of jovian broadband kilometric radiation ( bkom ) .
some observations favor the model in which bkom is produced due to the electron - cyclotron maser instability on open magnetic field lines at high magnetic latitudes @xcite .
however , this mechanism is difficult to reconcile with the observed fine spectral structure @xcite .
other studies have located the bkom sources in the io plasma torus , near the equatorial plane , where the emission is to be produced by a plasma mechanism @xcite .
this model seems to be more suitable for interpretation of radio bursts with a fine spectral and/or temporal structure ; however , the exact mechanism responsible for the formation of parallel stripes of emission in bkom spectra has yet to be investigated . the spectral structure shown in fig .
[ fig01 ] has a striking resemblance to the so - called zebra patterns that are often observed in the dynamic spectra of sporadic solar radio emission ( although at much higher frequencies : usually from @xmath6 mhz to a few ghz ) .
typical spectra of the solar zebra patterns are presented , e.g. , in the review of @xcite .
a number of models have been proposed to interpret these spectral structures
. therefore it would be interesting to apply these models to the jovian radio bursts .
in contrast to the highly dynamic conditions in solar flares , the magnetic field configuration and plasma distribution in the jovian magnetosphere are more stable and have been studied in situ , which provides more constraints on the radio emission mechanism .
zebra patterns are one of the most intriguing phenomena in solar radio astronomy ; more than ten theoretical models for their interpretation have been proposed to date ( see , e.g. , the reviews of * ? ? ?
* ; * ? ? ?
* where some of these models are discussed and compared with observations ) .
there is a general consensus that the emission is produced due to a plasma mechanism , because in the solar flaring loops the plasma frequency @xmath7 exceeds the electron cyclotron frequency @xmath8 ; here @xmath9 and @xmath10 are the electron number density and the magnetic field strength , respectively .
initially , zebra patterns were thought to be associated with excitation of harmonically - related plasma waves ( e.g. , bernstein modes ) ; however , such models can not account for the formation of non - equidistant zebra stripes .
therefore , the inhomogeneous models have been proposed in which the different zebra stripes are produced at different locations . at present ,
the most well - developed and observationally - supported model is the one based on the so - called double plasma resonance effect @xcite . in this model , it is assumed that an electron beam with unstable distribution ( e.g. , of the loss - cone type ) excites plasma ( upper - hybrid ) waves which are then transformed into electromagnetic ones due to nonlinear processes .
the generation efficiency of the plasma waves increases significantly if their frequency ( which is close to the upper - hybrid frequency @xmath11 ) coincides with a harmonic of the electron cyclotron frequency , i.e. @xmath12 where @xmath13 ( since @xmath14 , the double plasma resonance is possible only at @xmath15 ) . in an inhomogeneous coronal magnetic loop ( where the ratio @xmath16 is variable ) ,
condition ( [ dpr ] ) is satisfied at different heights for the different harmonic numbers , which results in the formation of the striped spectrum .
note that , although condition ( [ dpr ] ) includes the discrete harmonic number , the zebra stripes are not necessarily harmonically - related .
in contrast , the frequency intervals between the adjacent stripes can be variable and are determined only by the magnetic field and plasma density profiles ( i.e. , by the dependencies of these parameters on the coordinate along the magnetic loop ) . to form a distinctive zebra pattern , the emission has to be generated in a relatively narrow magnetic tube ; otherwise , due to transverse inhomogeneities of plasma and magnetic field
, the stripes would broaden and overlap .
although the exact distributions of the magnetic field and plasma in solar flares are unknown , the use of some reasonable assumptions ( e.g. , extrapolated photospheric magnetic fields ) allows us to reproduce the observed spectra of zebra patterns and locations of the sources of their stripes ( e.g. , * ? ? ?
in this paper , we adopt the above described double plasma resonance model .
we assume that the emission is generated on a given magnetic field line . at each point
, the emission is generated in a narrow range of frequencies near the local upper - hybrid frequency ; the total emission intensity at a given frequency @xmath17 is calculated as @xmath18 ^ 2}{\delta f^2(l)}\right\}\mathrm{d}l,\ ] ] where @xmath19 is the local emissivity , @xmath20 is the local emission bandwidth , and @xmath21 is the coordinate along the field line ( only the regions where @xmath14 are considered )
. the emissivity increases significantly if the double plasma resonance condition ( [ dpr ] ) is satisfied ; this effect is modeled by the expression @xmath22 ^ 2}{\delta s^2(l , s)}\right\},\ ] ] where @xmath23 is the parameter determining the modulation depth of the plasma emission mechanism . for simplicity , we assume that the relative bandwidth @xmath24 as well as the ratio @xmath25 are constant ( see also * ? ? ?
* ) . in all our calculations , we use @xmath26 , which is similar to the observed relative bandwidth of the zebra stripes ( see fig .
[ fig01 ] ) .
actually , the frequency of upper - hybrid waves and hence the radio emission frequency always slightly exceed the upper - hybrid frequency ; also , due to relativistic effects , the growth rate of upper - hybrid waves can reach its maximum when the ratio @xmath27 slightly differs from an integer @xcite
. however , these effects are negligible in comparison with the uncertainties in the plasma density and magnetic field models .
formulae ( [ mod1][mod2 ] ) do not allow us to calculate the actual emission intensity , because this requires knowing the parameters of energetic electrons and their distribution along the magnetic field line . also , we do not consider explicitly the process of transformation of the plasma waves into electromagnetic ones ; in contrast to the solar corona , this transformation in the jovian magnetosphere is more likely to be due to the linear conversion @xcite .
nevertheless , our model allows us to obtain a qualitative emission spectrum , i.e. , to find the frequencies where the zebra stripes can occur .
now we need to fill the emission model with the realistic magnetic field and plasma density profiles , i.e. , to calculate these parameters as the functions of the coordinate along a chosen magnetic field line . for the magnetic field
, we use the vip4 model @xcite . for the electron density ,
we use the analytical model proposed by @xcite .
this model describes the plasma density distribution in the plasma sheet of jupiter near the equatorial plane at the distances of @xmath28 from the rotation axis , i.e. , just in the region where bkom can be produced due to a plasma mechanism .
the model of @xcite is axisymmetric , which is obviously a simplification ; nevertheless , we believe that it provides a sufficiently good approximation to the actual plasma distributions . to account for the possible measurement uncertainties and long - term variations
, we also use the `` up - scaled '' models in which the density obtained from the formulae of @xcite is multiplied by a constant scaling factor in the range of @xmath29 . .
in panels ( a ) , ( b ) , and ( c ) , the plasma density is described by the model of @xcite scaled up by the factors of 1 , 2 , and 3 , respectively.,title="fig : " ] + . in panels ( a ) ,
( b ) , and ( c ) , the plasma density is described by the model of @xcite scaled up by the factors of 1 , 2 , and 3 , respectively.,title="fig : " ] + . in panels ( a ) ,
( b ) , and ( c ) , the plasma density is described by the model of @xcite scaled up by the factors of 1 , 2 , and 3 , respectively.,title="fig : " ]
the shape of the emission spectrum depends on the chosen `` active '' magnetic field line ( which is identified by the coordinates of the point where it intersects the centrifugal equatorial plane ) and the scaling factor applied to the plasma density model .
figure [ fig02 ] demonstrates the simulated spectra @xmath30 and the variation of these spectra with the radius of the chosen field line .
the field lines are assumed to intersect the equatorial plane at a system iii longitude of @xmath31 ; this longitude corresponds to the intersection of centrifugal and magnetic equators , so that the emissions produced in the northern and southern hemispheres have symmetric spectra .
one can see that the best agreement with the observations ( i.e. , five emission stripes in the frequency range of @xmath32 khz ) is achieved for the density scaling factor of about two ( fig . [
fig02]b ) and the `` active '' field line apex of about @xmath33 @xmath34 .
it is interesting to note that similar electron densities ( i.e. , twice as high as in the model of * ? ? ?
* ) have been inferred from the _ voyager 1 _ measurements in march 1979 @xcite .
the only disagreement with the observations is that the simulated frequency intervals between the adjacent zebra stripes become narrower with increasing emission frequency , while the observed intervals are nearly constant or even increase with the frequency ( see fig .
[ fig01 ] ) .
@xmath34 from the planet center ; the plasma density is described by the model of @xcite scaled up by the factor of 2 .
the emissions produced in different hemispheres are shown separately , in panels ( a ) and ( b ) .
vertical dashed lines correspond to system iii longitudes of @xmath31 and @xmath35.,title="fig : " ] + @xmath34 from the planet center ; the plasma density is described by the model of @xcite scaled up by the factor of 2 .
the emissions produced in different hemispheres are shown separately , in panels ( a ) and ( b ) .
vertical dashed lines correspond to system iii longitudes of @xmath31 and @xmath35.,title="fig : " ] ) .
thick solid lines are the magnetic field lines .
gray ( or color in the online version ) shades at @xmath36 @xmath34 represent the plasma density distribution , according to the model of @xcite scaled up by the factor of 2 ; the contours correspond to the plasma frequencies of @xmath37 , 100 , 30 , and 10 khz .
diamonds mark the sources of emission stripes , i.e. , the points where the double plasma resonance condition ( for the shown cyclotron harmonic numbers ) is satisfied . dashed lines bound the region where @xmath14 . ]
the exact longitude of the emission source in the considered event is unknown . to date ,
analysis of observations has found no indication of a preferable active longitude of bkom @xcite .
therefore we consider all possible source locations .
figure [ fig03 ] demonstrates the variation of the emission spectra with the longitude of the `` active '' magnetic field line ; the field lines are assumed to intersect the equatorial plane at a fixed distance from the planet center .
one can see that , except of the longitudes of @xmath38 and @xmath35 , the emission spectra produced in opposite hemispheres are clearly asymmetric .
the total emission ( i.e. , summed over both hemispheres ) can have rather complicated spectrum ; the emission stripes can merge and overlap .
therefore , to form a clear and regular zebra pattern , the emission should be generated either ( i ) near the above mentioned longitudes where the centrifugal and magnetic equators intersect ( as shown , e.g. , in fig .
[ fig02 ] ) or ( ii ) only in one hemisphere .
we think the latter variant is more likely because it allows us to reproduce better the observed frequency intervals between the adjacent zebra stripes : e.g. , in fig .
[ fig03]a ( northern hemisphere ) at the longitudes around @xmath39 , one can notice a slight increase of these intervals with the emission frequency which agrees very well with the observations .
the absence of emission from the opposite hemisphere can be caused , e.g. , by its absorption in the dense plasma torus @xcite .
the obtained agreement of the simulations and observations , however , is still qualitative , since a quantitative interpretation of the radio observations requires using more accurate ( and , possibly , non - stationary ) plasma and magnetic field models . the possible geometry of the radio emission source is shown in fig .
[ fig04 ] . as has been said above , the emission is generated near the equatorial plane ( at the latitudes of up to @xmath40 ) . for each magnetic field line ,
the points where the double plasma resonance condition ( [ dpr ] ) is satisfied are distributed over a sufficiently long distance up to several jovian radii .
the frequencies of the corresponding zebra stripes can be found from fig .
[ fig02 ] ( higher frequencies correspond to higher cyclotron harmonic numbers ) ; the highest frequencies are produced near the field line apex .
the formation of the emission stripes requires , in addition to satisfying the resonance condition , the presence of energetic electrons with an unstable distribution . according to fig .
[ fig04 ] , these electrons must fill ( more or less evenly ) a significant part of an `` active '' magnetic field line .
the radio burst shown in fig .
[ fig01 ] exhibits a complicated temporal behaviour , including fast frequency drifts : e.g. , during the time interval from 03:00:00 to 03:00:20 scet , the frequencies of the zebra stripes increase by about 20 khz .
the frequency drifts can have different origins ; in particular , the models proposed to explain the jovian decametric s- and l - bursts include propagation of spatially - localized electron beams through the magnetosphere , nonlinear wave - particle interactions , propagation of the radiation through a turbulent medium , etc .
( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) .
however , these models can not account for the observed parallel drift of the zebra stripes , because this would require highly synchronized simultaneous processes at different locations where the stripes seem to be produced .
as said above , in the double plasma resonance model the energetic electrons are assumed to be nearly omnipresent , while the emission stripes just mark the locations where the emission generation is more efficient . in the double plasma resonance model ,
frequency drifts can be caused by : ( i ) variations of the plasma density and/or magnetic field ; ( ii ) spatial movement of the electron acceleration site , which changes the currently `` active '' ( radio - emitting ) magnetic tube . in the former case
, the changes of the plasma density or magnetic field must take place simultaneously in the whole radio - emitting region ( see fig .
[ fig04 ] ) , in order to keep the zebra stripes nearly parallel . the above mentioned increase of the emission frequency
requires either an increase of the magnetic field by about 50% or even stronger increase of the plasma density ( see , e.g. , figs .
[ fig02]a-[fig02]c where the frequencies of the zebra stripes are almost insensitive to the density scaling factor ) .
we think that such rapid , strong , and global variations in the jovian magnetosphere are highly unlikely .
on the other hand , the observed frequency drifts can be easily explained by radial movement of the electron acceleration site . from fig .
[ fig02]b , one can find that the increase of the emission frequency by 20 khz corresponds , e.g. , to a decrease of the `` active '' magnetic field line radius from 11.4 to 10.0 @xmath34 .
the corresponding speed of the acceleration site ( if the site is located at the field line apex ) is about 5000 km @xmath41 , which is comparable to the electron thermal speed .
if the acceleration site is not located at the field line apex , its speed may be even lower .
azimuthal movement of the acceleration site is less likely : according to fig .
[ fig03 ] , a change of the emission frequency by 20 khz requires a change of the `` active '' longitude by @xmath42 .
note that the speed of the acceleration site is not related to the speed of the energetic electrons themselves .
we can only state that the electrons speed should be much higher than the above mentioned 5000 km @xmath41 to ensure a fast response of the emission spectrum to the movement of the acceleration site ; the electrons energy of a few kev seems to be well sufficient .
in the earlier part of the considered event , the emission spectrum looks completely different : there are relatively broadband pulsations with a period of about 2 s. most likely , these pulsations reflect the dynamics of the accelerated electrons , i.e. , a quasi - periodic regime of acceleration . in addition , the accelerated electrons occupy a rather wide range of magnetic field lines ( i.e. , the radio - emitting magnetic tube is relatively broad ) , so that the radio bursts produced at different double plasma resonance levels broaden and overlap in frequency , thus forming broadband pulsations instead of a zebra - like spectrum . later in the event ,
the electron acceleration regime becomes more stationary ( although one can still notice some pulsations around 03:00:00 scet ) and the acceleration region shrinks ( i.e. , the radio - emitting magnetic tube becomes narrower ) , so that the emission spectrum evolves into the zebra pattern .
a narrow radio - emitting magnetic tube is a key factor in the formation of zebra pattern : according to figs .
[ fig02 ] and [ fig03 ] , different zebra stripes will overlap if the sizes of the tube s cross - section ( in the equatorial plane ) exceed @xmath43 @xmath34 or @xmath44 in the radial or azimuthal directions , respectively ; clearly , the requirements to the radial extent are more severe .
we have investigated the formation of a zebra - like fine spectral structure in the jovian bkom .
it has been found that the emission model based on the double plasma resonance effect , together with the observation - based models of the magnetic field and plasma density in the jovian magnetosphere , allows us to reproduce the observed spectra .
the observed frequency drifts , most likely , reflect the dynamics of the source of the accelerated electrons ( this conclusion may be valid also for the solar radio bursts ) .
zebra patterns can be formed only if the electron acceleration site is very compact , so that the magnetic tube occupied by these electrons is very narrow .
since the bursts with zebra patterns are uncommon , this condition seems to be rarely met .
more extended emission sources ( where the emissions from different double plasma resonance levels overlap in frequency ) should produce more or less continuous broadband spectra that are typical of bkom .
radio bursts with zebra patterns can potentially be used as a diagnostic tool for the jovian magnetosphere ; e.g. , they could allow us to reconstruct the plasma density and magnetic field profiles
. however , such work requires more comprehensive observations , including an accurate determination of the emission source location .
a.a . kuznetsov thanks the leverhulme trust for financial support .
research at armagh observatory is grant - aided by the northern ireland department of culture , arts and leisure .
this work was supported in part by the russian foundation for basic research , grant no .
12 - 02 - 00173 .
the authors thank bill kurth for providing the _ cassini _ data and fran bagenal for her help with the plasma density model .
32 natexlab#1#1url # 1`#1`urlprefix[2][]#2 [ 2]#2 , . .
, , , , . . ,
. . , . , , , ,
. . , . , , ,
. . , . , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,
. . , . , . , in : , , ( eds . ) , , pp . . , , . .
, . , , , , , .
. . , . , , , , ,
. . , . , , , , , . , in : , , ( eds . ) , , p. . , , , . .
, , , , , , , , . , in : , , ( eds . ) , , p. . , , . .
, , , . . ,
, in : , , ( eds . ) , , pp . . , , a. . , . , , b. , in : , , ( eds . ) , , p. .
, , , , , , . . ,
, , , , , , , , , , a. . , .
, , , , , , , , , , , , b. . , . , . .
, , , . . ,
, , , . . , | we investigate the formation of zebra - like fine spectral structures ( consisting of several parallel bands in the dynamic spectrum ) in the jovian broadband kilometric radiation ; such radio bursts were observed by _
cassini _ in 2000/2001 .
we assume that the emission is generated due to a plasma mechanism in the io plasma torus .
we have shown that the double plasma resonance effect ( that was proposed earlier as a formation mechanism of the solar zebra patterns ) is able to produce the observed spectral structures .
the observed frequency drifts are caused , most likely , by the dynamics of the electron acceleration site .
the required conditions in the emission source are discussed . zebra pattern , double plasma resonance , jupiter magnetosphere , kilometric radio emission |
the @xmath0 reaction , which is one of the reactions with four charged particles in the final state , can be easily measured . very close to the threshold
the excitation of the roper resonance and its subsequent decay as well as double-@xmath2 excitation constitute the dominant reaction mechanism @xcite . only energy dependence of the total cross section was discussed in @xcite . the pion - pion rescattering mechanism shown in fig.[fig : pion - pion - rescattering ]
was not discussed so far in the literature .
the pion - pion rescattering mechanisms of exclusive production of @xmath4 and @xmath5 in proton - proton and proton - antiproton collisions .
some kinematical variables are shown explicitly .
the stars attached to @xmath6 mesons denote the fact that they are off - mass - shell .
, scaledwidth=40.0% ] on the other hand a significant progress in studying pion - pion scattering at low - energy has been recently achieved due to works based on dispersive analyses of experimental data @xcite .
these works were sometimes used together with chiral perturbation theory ( chpt ) to fix amplitude @xcite and sometimes to test chpt predictions @xcite .
they led to precise determination of the @xmath7 scattering amplitudes consistent with analyticity , unitarity and crossing symmetry .
strong theoretical constraints from forward dispersion relations and sum rules allowed to test several , sometimes conflicting sets of data @xcite . together with twice subtracted dispersion relations ( the roy s equations ) these constraints helped to significantly diminish the errors of , for example , sigma pole position and @xmath8-wave scattering lengths @xcite .
recent works on once subtracted dispersion relations confirmed these results @xcite .
application of roy s equations in another dispersive analysis of experimental data allowed to eliminate the long standing `` up - down '' ambiguity below 1 gev in @xmath9 is angular momentum between pions and @xmath10 is the total isospin of the pion pair . ]
= @xmath11 wave @xcite .
the simple and model independent amplitudes of the @xmath11 , @xmath12 , @xmath13 , @xmath14 , @xmath15 , @xmath16 and @xmath17 waves presented in series of works @xcite seem to be efficient and easy to use in analyses of @xmath7 interactions .
amplitudes presented in @xcite have been applied in this analysis to parameterize the final state interactions @xmath18 .
the knowledge from the @xmath19 reaction can almost directly be used in our @xmath20 reaction shown in fig.[fig : pion - pion - rescattering ] .
it is the aim of this paper to evaluate the pion - pion rescattering contribution for the measured close - to - threshold region of the @xmath0 reaction as well as to make predictions for the future experiments with the panda detector at high energy storage ring ( hesr ) in gsi darmstadt for the @xmath1 reaction .
it is straightforward to evaluate the pion - pion exchange current contribution shown in fig.[fig : pion - pion - rescattering ] .
if we assume the @xmath21 type coupling of the pion to the nucleon then the born amplitude squared and averaged over initial and summed over final spin polarizations reads : @xmath22 \times 2 \nonumber \\
\times & \dfrac{g_{\pi nn}^2}{(t_1 - m_{\pi}^2)^2 } f_{\pi nn}^2(t_1 ) \ ; \times \ ; |{\cal m}_{\pi^{0*}\pi^{0*}\to\pi^+\pi^-}(s_{34},cos\theta^ { * } , t_{1},t_{2})|^2 \ ; \times \ ; \dfrac{g_{\pi nn}^2}{(t_2 - m_{\pi}^2)^2 } f_{\pi nn}^2(t_2 ) \nonumber \\ \times & \left [ \left ( e_b + m \right)\left ( e_2 + m \right )
\left(\dfrac{{\bf p}_b^2}{(e_2 + m)^2 } + \dfrac{{\bf p}_2 ^ 2}{(e_2 + m)^2 } - \dfrac { 2 { \bf p}_b \cdot { \bf p}_2}{(e_b + m)(e_2 + m)}\right)\right ] \times 2 \ ; .
\label{pion - pion_amplitude}\end{aligned}\ ] ] in the formula above @xmath23 is the mass of the nucleon , @xmath24 and @xmath25 are the energies of initial and outgoing nucleons , @xmath26 and @xmath27 are the corresponding three - momenta and @xmath28 is the pion mass .
the factor @xmath29 is the familiar pion nucleon coupling constant and is relatively well known @xcite ( @xmath30 = 13.5 14.6 ) . in the case of central production of pion pairs
not far from the threshold region rather large transfer momenta squared @xmath31 and @xmath32 are involved and one has to include non - point - like and off - shellness nature of the particles involved in corresponding vertices .
this is incorporated via the @xmath33 or @xmath34 vertex form factors .
we shall discuss how the uncertainties of the form factors influence our final results . in the meson exchange approach
@xcite they are parameterized in the monopole form as @xmath35 in the following for brevity we shall use notation @xmath36 which means @xmath31 or @xmath32 .
typical values of the form factor parameters are @xmath37 = 1.21.4 gev @xcite , however the gottfried sum rule violation prefers smaller @xmath38 0.8 gev @xcite .
the amplitude of the subprocess @xmath39 with virtual initial pions is written in terms of the amplitude for real intial pions and correction factors as : @xmath40 the on - shell amplitude can be expanded into partial wave amplitudes @xmath41 with angular momentum @xmath42 and isospin @xmath10 : @xmath43 for a limited range of @xmath44 it is enough to take only a few partial waves . in our calculation
@xmath41 can be parameterized in terms of the pion - pion phase shifts @xmath45 and the inelasticities @xmath46 taken from @xcite @xmath47 in the formula above @xmath48 are extra correction factors due to off - shellness of initial pions .
we use exponential form factors of the type @xmath49 i.e. normalized to unity on the pion - mass - shell . in general , the parameter @xmath50 is not known but in principle could be fitted to the experimental data providing that our mechanism is the dominant mechanism which can be true only in a limited corner of the phase space . from our general experience in hadronic physics
we expect @xmath51 1 gev .
the @xmath52 in eq.(2.4 ) requires a seperate discussion . in the on - shell case
the @xmath53 can be expressed in terms of the two - body mandelstam invariants @xmath54 and @xmath55 in two equivalent ways : @xmath56 this can be generalized to the case of off - shell initial pions as : @xmath57 in our case of the @xmath58 reaction reaction denotes a type of the reaction with two initial and four final particles .
] we have to deal with off - shell initial pions and an analytical continuation of formula ( [ costheta_off - shell_1 ] ) is required . in the following we use the most straightforward prescription : @xmath59 the formula above reproduces the on - shell formula ( [ costheta_on - shell ] ) when @xmath60 and @xmath61 , is symmetric with respect to @xmath54 and @xmath55 and fulfils the requirement -1 @xmath62 1 .
the differential cross sections for the @xmath58 reaction are calculated using the general formula @xmath63
close to the two - pions production threshold the dominant mechanism for the reaction @xmath0 is roper resonance excitation and its subsequent three - body decay @xcite .
this mechanism constitutes an `` unwanted background '' to our pion - pion rescattering . at low energy the sigma and
pion exchanges are the dominant mechanisms of roper resonance excitation ( see @xcite ) . here
we show how to approximately estimate the phase - space integrated contribution of the mechanism shown in fig.[fig : roper_mechanism ] not very close to the threshold . final state ( see @xcite ) . ]
the dominant mechanisms of roper resonance production at low energy proton - proton scattering .
, title="fig:",scaledwidth=45.0% ] the dominant mechanisms of roper resonance production at low energy proton - proton scattering .
, title="fig:",scaledwidth=45.0% ] the amplitude for the roper resonance @xmath64 excitation via @xmath65-meson exchange can be written as @xmath66 \frac{1}{t - m_{\sigma}^2 } g_{\sigma n n^ * } f_{\sigma
n n^*}(t ) [ { \bar u}(p_{n^*},\lambda_{n^ * } ) u(p_b,\lambda_b ) ] \ ; , \nonumber \\ { \cal m}_{\lambda_a \lambda_b \to \lambda_n \lambda_{n^*}}^{(\sigma , u ) } & = & g_{\sigma n n^ { * } } f_{\sigma n n^*}(u ) [ { \bar u}(p_{n^*},\lambda_{n^ * } ) u(p_a,\lambda_a ) ] \frac{1}{u - m_{\sigma}^2 } g_{\sigma n n } f_{\sigma n n}(u ) [ { \bar u}(p_n,\lambda_n ) u(p_b,\lambda_b ) ] \;.\nonumber \\ \label{roper_excitation_sigma_amplitude}\end{aligned}\ ] ] the amplitude for the roper resonance excitation via @xmath67-exchange mechanism can be written as @xmath68 \frac{1}{t - m_{\pi}^2 } \nonumber \\ & & g_{\pi n n^ * } f_{\pi
n n^*}(t ) [ { \bar u}(p_{n^*},\lambda_{n^ * } ) i \gamma_5 u(p_b,\lambda_b)]\ ; , \nonumber \\ { \cal m}_{\lambda_a \lambda_b \to \lambda_n \lambda_{n^*}}^{(\pi , u ) } &
= & g_{\pi n n^ { * } }
n n^{*}}(u ) [ { \bar u}(p_{n^*},\lambda_{n^ * } ) i \gamma_5 u(p_a,\lambda_a ) ] \frac{1}{u - m_{\pi}^2 } \nonumber \\ & & g_{\pi n n } f_{\pi n n}(u ) [ { \bar u}(p_n,\lambda_n ) i \gamma_5 u(p_b,\lambda_b)]\;. \label{roper_excitation_pion_amplitude}\end{aligned}\ ] ] in the above equations @xmath69 , @xmath70 , @xmath71
, @xmath72 represent the coupling constants and @xmath73 denotes proton or antiproton ; @xmath74 , @xmath75 , @xmath76 , @xmath77 , are the spionors of the protons and roper resonance ; @xmath78 and @xmath79 denote the four - momenta of the outgoing proton and the roper resonance ; @xmath80 and @xmath81 the helicities of the nucleon and the roper resonance ; @xmath82 and @xmath83 are the four - momentum transfers ; @xmath28 and @xmath84 are the mass of the pion and sigma mesons . in our calculations the coupling constants are taken as @xmath85 = 13.6 @xcite , @xmath86 = 5.69 @xcite , @xmath87 = 2.0 and @xmath88 = 2.0 . because numerically the @xmath65-exchange is the dominant mechanism and the @xmath67-exchange is only a small correction ,- exchange or pseudoscalar @xmath89 coupling for pion exchange . ] in practice the latter can be neglected .
the coupling constant @xmath72 is in fact an unknown parameter which in principle should be determined from the experimental data .
different values have been used in the literature @xcite .
our number is an average value of those found in the literature .
we parameterize the form factors @xmath90 ( and @xmath91 ) either in the monopole form with cut - off parameter @xmath92 as traditionally for low energy processes : @xmath93 or in the exponential form often used at high energies with cut - off parameter @xmath94 : @xmath95 the angular distribution for single roper resonance excitation can be calculated from the amplitude above as @xmath96 where @xmath97 is a square of the proton - proton center - of - mass energy ; @xmath98 and @xmath99 are center - of - mass momenta in the initial @xmath100 or the final @xmath101 systems , respectively and @xmath102 , where @xmath103 is the center - of - mass angle between the outgoing and initial nucleon .
the factor @xmath104 and @xmath105 emerge for the simple reason that the polarization of initial and final particles is not considered .
in general , one should calculate the cross section for 2 @xmath106 4 reaction based on diagrams shown in fig.[fig : roper_mechanism ] with roper resonance in the intermediate state ( in general off - shell particle ) .
however , for sufficiently high energies the total cross section for the @xmath3 final state can be written approximately as a cross section for the roper resonance excitation and a probability for the @xmath107 decay ( on - shell approximation ) : @xmath108 this formula will be used to calculate the total cross section for the roper resonance mechanism to show as a reference for the discussed above two - pion rescattering contribution .
the branching ratio to the @xmath109 channel is not very well known and the mechanism of the roper resonance decay can be complicated .
particle data book contains only branching fraction for all @xmath110 states .
our decay channel ( @xmath109 ) is only one out of three possible ( @xmath111 , @xmath109 , @xmath112 ) .
we take @xmath113 = 0.1 . in the next section
we shall show our predictions for several differential distributions in different variables .
before we go to our four - body reaction let us stay for a moment with the @xmath114 on - shell scattering . in fig.[fig
: pi0pi0_pippim ] we show the total ( angle - integrated ) cross section for the @xmath115 process which constitutes the subprocess in the @xmath58 reactions discussed in the present paper . here the partial wave expansion ( [ pipi_pipi_on - shell - amplitude ] ) with @xmath116 and @xmath117 parameterizations from ref.@xcite were used . in the present work we have limited to @xmath118 1.5 gev , i.e. energies relevant for wasa and future panda experiments . the angle - integrated cross section for the reaction @xmath114 .
the thick solid line represents the coherent sum of all partial waves .
the contributions for individual partial waves @xmath11 , @xmath13 , @xmath14 and @xmath15 are shown for comparison .
, scaledwidth=50.0% ] at higher @xmath119 larger @xmath120 energies may contribute .
this will be discussed elsewhere @xcite .
we show also individual contributions of different partial waves : ( @xmath121)=(0,0),(0,2),(2,0 ) and ( 2,2 ) . because of identity of particles in the initial state only partial waves with
even @xmath42 contribute .
one can see characteristic bumps related to the famous scalar - isoscalar @xmath65-meson at @xmath122 0.5 - 0.6 gev and the tensor - isoscalar @xmath123 .
the dip at @xmath124 980 mev is due to interference of the @xmath65 with another scalar - isoscalar narrow @xmath125 meson .
generally the contributions of nonresonant partial waves with @xmath10=2 are much smaller than those for @xmath10=0 . the phase - space integrated cross section for the @xmath0 reaction as a function of the proton kinetic energy in the laboratory frame @xmath126 together with the experimental data from refs.@xcite .
the thick solid line is explained in the text .
the uncertainties band is also shown . in all cases a coherent sum of all partial waves is taken.,scaledwidth=50.0% ] in fig.[fig : tp ] we show the proton energy excitation function of the integral cross section for the @xmath0 reaction . in addition , we compare our results with the experimental data for the @xmath0 reaction ( from refs.@xcite ) . and for the @xmath1 one ( only data from the jetset ( ps202 ) experiment at lear @xcite ) .
we present previous data ( open symbols ) with low statistics coming mainly from bubble - chamber measurements on hydrogen or on deuterium from refs.@xcite as well as one datum point from an inclusive spectrometer measurement at 800 mev @xcite .
the newer data taken from refs.@xcite ( full symbols ) are much closer to the threshold of the reaction and are an order of magnitude smaller .
we show how the uncertainties of the form factor parameters @xmath37 affect our final results .
for the pion - pion rescattering we modify the cut - off parameter @xmath37 in eq.(2.2 ) @xmath127 ( 0.8 , 1.4 ) gev ) and the cut - off parameter @xmath50 in eq.(2.6 ) @xmath128 ( 0.5 , 2.0 ) gev ) .
the thick solid line show theoretical predictions from the model calculations with @xmath129 gev and @xmath130 gev . the pion - pion rescattering contribution is found to be negligible .
as discussed in the theory section , the @xmath114 amplitude used for the @xmath114 reaction can , after a small `` correction '' for virtualities of both initial @xmath6 s , be used for the four - body process of our main interest . in fig.[fig : sig_tot_w ] we show the total cross section ( integrated over the whole phase space with the restriction @xmath131 1.42 gev ) for the four - body reaction as a function of the overall center - of - mass energy @xmath119 .
we show the coherent sum of partial waves for different @xmath42 and @xmath10 as well as individual contributions .
the maximum of the cross section occurs at @xmath132 5 gev , i.e. at the highest energy planned for the fair hesr .
the @xmath42 = 0 , @xmath10 = 0 partial wave has the dominant contribution .
the sum of the individual contributions is not equal to the cross section calculated with the sum of the partial wave amplitudes because of relatively strong interference effects .
the phase - space integrated cross section for the reaction @xmath1 as a function of center of mass energy @xmath119 .
the thick solid line represents the coherent sum of all partial waves .
the contributions for individual partial waves @xmath11 , @xmath13 , @xmath14 and @xmath15 are shown for comparison .
, scaledwidth=50.0% ] in fig.[fig : roper_w_minus ] we compare the pion - pion rescattering contribution and the contribution of roper resonance excitation through @xmath65-meson exchange . in both cases
we have estimated the uncertainties of the contributions .
for the pion - pion rescattering we modify @xmath37 in eq.(2.2 ) ( @xmath133 gev ) and @xmath50 in eq.(2.6 ) ( @xmath134 gev ) .
the bottom dashed line was obtained with @xmath129 gev and @xmath135 gev while the top dashed line with @xmath136 gev and @xmath137 gev . for the contribution of the roper resonance excitation through @xmath65-meson exchange
we modify @xmath138 gev ( band with vertical lines ) in the monopole parameterization and @xmath139 gev ( band with horizontal lines ) in the exponential parameterization . the phase - space integrated cross section for the reaction @xmath1 as a function of center of mass energy @xmath119 .
we compare the pion - pion rescattering contribution and the roper resonance contribution ( only @xmath65-meson exchange included ) .
the uncertainty bands for both contributions are also shown .
the area of uncertainties for the pion - pion rescattering contribution is indicated by the dashed lines .
pion rescattering contribution is a coherent sum of all partial waves .
, scaledwidth=50.0% ] because at low energies the roper resonance excitation and double-@xmath2 excitation play the dominant role it is not obvious how to extract the pion - pion rescattering contributions . to cut off the roper resonance excitation contribution we eliminate from the phase space those cases when : + @xmath140 or @xmath141 .
+ to suppress the double-@xmath2 excitation we eliminate from the phase space those cases when : + @xmath142 and @xmath143 or + @xmath144 and @xmath145 .
+ above @xmath146 and @xmath147 represent effective mass of the @xmath148 and @xmath149 systems , respectively ; @xmath150 and @xmath151 are cut - off parameters .
we take @xmath150 = 0.4 gev and @xmath151 = 0.2 gev which are considerably bigger than the decay widths . in fig.[fig : dsig_dt ] we present differential cross section @xmath152 ( integrated over all other variables ) for pion - pion rescattering only .
the shape of the distribution reflects tensorial structure of the @xmath153 vertices ( see eq.([pion - pion_amplitude ] ) ) as well as @xmath31 or @xmath32 dependence of vertex form factor ( see eq.([f_pinn_formfactor ] ) ) .
this plot illustrates how virtual are `` initial '' pions . in principle , measuring such distributions would allow to limit , or even extract , the @xmath153 form factor in relatively broad range of @xmath31 or @xmath32 .
this is not possible in elastic nucleon - nucleon scattering where many different exchange processes contribute .
differential cross section @xmath152 for the @xmath1 reaction at @xmath119 = 5.5 gev .
the solid line is the cross section without cuts , the dashed line includes cuts to remove regions of roper resonance and double-@xmath2 excitations . , scaledwidth=45.0% ] in fig.[fig : dsig_dptsum ] we show differential cross section @xmath154 , where @xmath155 . for collinear ( parallel to the parent nucleons )
initial pions this distribution would be proportional to the dirac @xmath156 .
the deviation from @xmath156 is therefore a measure of noncollinearity and is strongly related to virtualities of `` initial '' pions ( see fig.[fig : dsig_dt ] ) .
differential cross section @xmath154 for the @xmath1 reaction at @xmath119 = 5.5 gev .
the solid line is the cross section without cuts , the dashed line includes cuts to remove regions of roper resonance and double-@xmath2 excitations .
, scaledwidth=45.0% ] the two - pion invariant mass distribution given by the differential cross section @xmath157 is particularly interesting . here
( see fig.[fig : dsig_dmpipi ] ) one can see two characteristic bumps corresponding to the famous scalar - isoscalar @xmath158 meson and tensor - isoscalar @xmath123 meson as well as the dip from @xmath125 .
the cuts to remove regions of roper resonance and double-@xmath2 excitation only slightly modify the spectral shape .
differential cross section @xmath157 for the @xmath1 reaction at @xmath119 = 5.5 gev .
the solid line is the cross section without cuts , the dashed line includes cuts to remove regions of roper resonance and double-@xmath2 excitations . , scaledwidth=45.0% ] the panda detector is supposed to be a @xmath159 solid angle detector with good particle identification for charged particles and photons .
this opens a possibility to study several correlation observables for outgoing particles .
one of them is azimuthal angle correlation between charged outgoing pions @xmath160 , never discussed in the literature . in fig.[fig : dsig_dphi34 ] we present differential cross section @xmath161 .
clearly a preference of back - to - back emissions can be seen .
imposing cuts on the roper resonance and double-@xmath2 excitation lowers the cross section but only mildly modifies the shape .
because the shape of the azimuthal angle correlations strongly depends on the reaction mechanism , measuring such correlation would provide then very valuable information .
differential cross section @xmath162 for the @xmath1 reaction at @xmath119 = 5.5 gev .
the solid line is the cross section without cuts , the dashed line includes cuts to remove regions of roper resonance and double-@xmath2 excitations .
, scaledwidth=45.0% ] in fig.[fig : dsig_dy3dy4 ] we show differential cross section @xmath163 in the two - dimensional space @xmath164 . for comparison in the right panel
we show a similar distribution when extra cuts to remove regions of roper resonance and double-@xmath2 excitation are imposed .
the cuts do not much affect the region of @xmath165 0 . in practice , the cuts on the roper resonance region do not modify the results .
the cuts on roper resonance act for @xmath166 and @xmath167 or @xmath168 and @xmath169 i.e. in the region where the two - pion rescattering contribution is small .
the cuts on double-@xmath2 excitation act for @xmath166 and @xmath169 or @xmath168 and @xmath167 i.e. in the region where the two - pion rescattering contribution is sizeable .
this shows that the double-@xmath2 excitation is more critical than the roper resonance excitation in the context of extracting the pion - pion rescattering contribution .
two - dimensional differential cross section @xmath170 in @xmath171 for the @xmath1 reaction at @xmath119 = 5.5 gev ( left panel ) . in the right panel
we have included in addition cuts to remove regions of roper resonance and double-@xmath2 excitations . ,
title="fig:",scaledwidth=45.0% ] two - dimensional differential cross section @xmath170 in @xmath171 for the @xmath1 reaction at @xmath119 = 5.5 gev ( left panel ) . in the right panel
we have included in addition cuts to remove regions of roper resonance and double-@xmath2 excitations .
, title="fig:",scaledwidth=45.0% ]
we have calculated both differential and total cross sections for the @xmath0 and @xmath1 reactions close to threshold and for future panda experiments .
our results have been compared with very close to threshold data measured by the wasa collaboration .
we have shown that very close to threshold the pion - pion rescattering mechanism gives much smaller contribution than the excitation of the roper resonance and its subsequent decay as well as the double-@xmath2 excitation and subsequent decays studied in the past @xcite . at low energies all these mechanisms overlap and it is not possible to extract the pion - pion rescattering contributions and therefore not possible to study the @xmath114 process .
going to higher energies allows to find regions of the final state phase space where the pion - pion rescattering process dominates over the roper resonance and double-@xmath2 excitation mechanisms .
experiments at highest energies of the hesr ( fair project ) at gsi darmstadt open a possibility to study the pion rescattering process and provide an excellent place for studying properties of the roper @xmath172 resonance .
we have presented several distributions which could be measured in the future with the panda detector at the gsi hesr . particularly interesting
is the distribution in two - pion invariant mass , where one should observe bumps related to the famous scalar - isoscalar @xmath65-meson and to tensor - isoscalar @xmath123 meson as well as a dip from the interference with @xmath125 and @xmath65 .
this distribution is slightly different compared to the dependence of the total @xmath114 cross section on @xmath120 .
this is caused mainly by the four - body phase space modifications .
the pions from the pion - pion rescattering are produced preferentially in opposite hemispheres , i.e. if one pion is produced at positive center - of - mass rapidities the second pion is produced at negative ones .
this is similar to the double-@xmath2 excitation mechanism .
imposing cuts on double-@xmath2 excitation leaves untouched the region of midrapidities .
also the region of large @xmath173 stays unmodified by the cuts on double-@xmath2 excitations .
* acknowledgments * discussions and communications with benoit loiseau are acknowledged .
this work was partially supported by the polish ministry of science and higher education ( mnisw ) under contract mnisw n n202 249235 . | we evaluate the contribution of pion - pion rescattering to the @xmath0 and @xmath1 reactions .
we compare our results with the close - to - threshold experimental data .
the pion - pion rescattering contribution is found there to be negligible .
the predictions for future experiments with panda detector at high energy storage ring ( hesr ) in gsi darmstadt are presented .
it is discussed how to cut off the dominant roper resonance and double-@xmath2 excitation mechanisms leading to the @xmath3 channel in final state .
differential distributions in momentum transfers , transverse momentum , effective two - pions mass , relative azimuthal angle between pions , and pion rapidities are presented . |
fu orionis systems are a class of exceptionally luminous young stellar objects found in star - forming regions ( hartmann & kenyon 1996 ) . originally identified from their very large increases in optical brightness over timescales of years or less ( herbig 1977 ) , a larger group of heavily extincted probable members of the class have been identified by their characteristic infrared spectra , which strongly differ from those of typical t tauri stars ( reipurth & aspin 1997 ; aspin & reipurth 2003 ; reipurth et al .
additional support for the identification of these heavily embedded objects comes from recent high - resolution infrared spectroscopy , which indicates that many of these sources are rapidly rotating and exhibit double - peaked absorption line profiles as observed in fu ori ( greene , aspin , & reipurth 2008 ) . the accretion disk model for fu ori objects ( hartmann & kenyon 1996 , and references therein ) rests fundamentally on the need to explain the peculiar spectral energy distributions ( seds ) of these objects , which are much broader than that of a single temperature blackbody or star , and which exhibit a continuously varying spectral type as a function of wavelength .
the disk model naturally accounts for these properties , as observations at longer wavelengths probe increasingly cooler disk regions with later spectral types ; our detailed model for fu ori matches the sed from optical wavelengths to the mid - infrared region ( zhu et al .
2007 , 2008 ) .
in addition , the disk model predicts that differential rotation should be observed , with slower rotation seen at longer wavelengths , which arise from outer disk radii , and this has been confirmed by comparing optical ( @xmath5 m ) and near - infrared ( @xmath0 m ) spectral line profiles ( hartmann & kenyon 1987a , b ; kenyon , hartmann , & hewett 1988 ) .
a desirable feature in a model or theory is an ability to make predictions that can be tested observationally .
the disk model predicts that the observed rotational spectral line broadening should be even smaller at @xmath6 m ( using the fundamental co vibrational transitions ) than at @xmath7 m ( using the first overtone co vibrational lines ) . in this paper
we present a high - resolution spectrum of fu ori in the @xmath8 m region which matches the predictions of the disk model .
we also show that some discrepancies seen in optical spectra of fu ori in comparison with simple disk models are alleviated by a decrease in the maximum disk temperature which we proposed in zhu et al .
a high - resolution spectrum of fu ori at @xmath9 m was obtained at ut 01:00:29 on 2007 february 4 using the phoenix spectrometer ( hinkle et al .
1998 , 2000 , 2003 ) on the 8-m gemini south telescope .
observations were taken with a two - pixel slit ( 0.17@xmath10@xmath10 ) for a resolution of @xmath11 over the wavelength range @xmath12 m .
we observed fu ori at two positions along the slit for eight 2 minute exposures .
we also observed the b2 iii star hr1790 for telluric line correction and took 10 flat - field and dark images .
we reduced the data using iraf .
we averaged the flat - field and dark images and subtracted the average dark image from the average flat - field image .
this averaged , dark - subtracted flat - field image was then divided into the target spectra .
images at different positions of the slit were differenced to remove the sky and dark backgrounds .
we then extracted the spectra using the iraf @xmath13 routine and later combined and flattened the spectra using @xmath14 in iraf .
the spectrum of the hot star hr1790 was used to divide out the telluric lines from the fu ori spectrum .
wavelength calibration was computed using telluric lines from the arcturus atlas of hinkle et al .
( 1995 ) .
we also obtained a high - resolution spectrum of fu ori at ut 17:59:15 on 2008 november 14 using the magellan inamori kyocera echelle ( mike ) spectrograph on the 6.5 m magellan clay telescope at las campanas observatory ( bernstein et al .
mike is a double echelle spectrograph which delivers full wavelength coverage from about @xmath15 ( blue side ) and @xmath16 ( red side ) .
the data were obtained in subarcsecond seeing with a 0.7slit , binning 2x2 and an exposure time of 120s .
the resolutions were @xmath4 40,000 and @xmath17 30,000 for the blue and red sides respectively .
the mike data were reduced using the mike redux idl pipeline.burles/www/mike/ ]
figure 1 shows the reduced phoenix spectrum of fu ori .
strong telluric features mean that regions around wavenumber 2009.3 @xmath18 are not usable , and small residuals from the telluric correction can be seen at wavenumbers 2010.4 , 2010.9 , 2011.9 , and 2013.5 @xmath18 .
the spike around wavenumber 2012.5 @xmath18 is due to a bad pixel . outside of these regions ,
an absorption spectrum is clearly present , with relatively broad features and substantial blending .
most of the lines are due to the p - branch of the fundamental rotational - vibrational transitions of co. to interpret the results , we calculated a synthetic disk spectrum using the methods described by zhu et al .
the model parameters were those used by zhu et al .
( 2007 ) to fit the sed of fu ori and to match the observed rotational broadening observed at optical and near - infrared wavelengths : central star mass @xmath4 0.3 m@xmath19 , mass accretion rate of the inner high @xmath20 disk @xmath42.4@xmath2110@xmath22 , disk inner radius @xmath45 @xmath23 , outer radius of the inner high @xmath20 disk @xmath4 1 au , and the disk inclination angle @xmath455@xmath24 .
( see figure 8 of zhu et al . for the fit to the @xmath7 m co lines ) .
we emphasize that we have _ not _ changed or adjusted any parameters from the zhu et al .
( 2007 ) fu ori model ; these are predicted spectra . the lower dotted curve in figure 1 shows the synthetic disk spectrum observed pole - on , so that individual spectral lines can be seen without the blending that occurs due to the large rotational broadening .
comparison of the nonrotating spectrum with the observations shows that fu ori has significantly larger line widths , consistent with rapid rotation , and unlike profiles of m giants and supergiants .
the upper dotted curve shows the synthesized spectrum using the inclination and central mass used to obtain a match to the @xmath7 m co line widths .
the agreement between synthetic and real spectra is quite good , except near @xmath25 where we are missing the co 5 - 4 p7 line in the model .
this could be due to adopting too small an oscillator strength .
the half - width at half - depth ( hwhd ) of the lines in this spectral region ( 5 @xmath26 m ) is @xmath27 , considerably smaller than the line widths measured at @xmath7 m ( hwhd @xmath4 36 km s@xmath28 ) ( hartmann & kenyon 1987a ; hartmann , hinkle , & calvet 2004 ; zhu et al .
this hwhd at 5 @xmath26 m is close to the keplerian velocity at 0.5 au around 0.3 m@xmath19 central star .
all the strong lines have been identified in the pole - on model spectrum .
though some water lines are present in this spectrum , they are washed out or blended in the broadened spectrum ( upper dotted curve ) .
only the strong co fundamental lines can be identified with certainty .
it is also worth noting that some of the unblended line profiles exhibit evidence for double - peaked shapes predicted by simple disk models ( e.g. , hartmann & kenyon 1987a , b ) .
in addition , some of the blends show sharp features ( eg . 2011.2 cm@xmath28 line ) which are the result of overlapping double - peaked lines .
these features naturally arise in a disk model but would not be seen in rotating star models ( unless large polar spots are invoked ; see below ) . in figure 2
we display a segment of the mike spectrum of fu ori in the wavelength range @xmath30 for comparison with the synthetic keplerian disk spectrum .
we again find good agreement between model and observation , demonstrating that there has been no change in the estimated optical rotational velocity of the object between the observations in zhu et al .
2007 , which we used to set the disk parameters , and this paper .
the hwhd of the optical lines in this wavelength range is @xmath465@xmath315 km s@xmath28 , consistent with hwhd @xmath4 62@xmath315 km s@xmath28 measured by petrov & herbig ( 2008 ; ph08 ) .
thus , compared with hwhd of 2 micron co first - overtone lines @xmath4 36@xmath313 km s@xmath28 and hwhd of 5 micron co fundamental lines @xmath4 22@xmath312 km s@xmath28 , the differential rotation in fu ori observed over nearly an order of magnitude in wavelength is consistent with keplerian rotation . the slow rotation observed at @xmath8 m implies spectral formation at radii out to @xmath32 au , in agreement with our sed modeling ; this supports our conclusion in zhu et al .
( 2007 , 2008 ) that the extent of the high - accretion rate disk is larger than can be explained by pure thermal instability models for outbursts @xcite .
the consistency of the variation of rotational velocities as observed over to @xmath33 m with keplerian rotation seemingly provides strong evidence for the accretion disk interpretation of fu ori .
however , ph08 recently argued that while the infrared spectrum is that of an accretion disk , the optical spectrum is produced by a rapidly - rotating star with a dark polar spot .
the ph08 argument against a pure disk model for a central star rests on three main points : there is no evidence for a variation of absorption line width as a function of excitation potential over the wavelength range @xmath34 m ; there is no evidence for a variation of rotational velocity with wavelength over that wavelength range ; and the observed line profiles are more `` boxy '' or flat - bottomed than the double - peaked profiles of the disk model .
it is important to recognize that the above - listed effects expected for a disk spectrum require not only differential rotation but a temperature gradient as well .
a keplerian disk exhibiting a constant effective temperature would not show any effect of rotational velocity with either excitation or wavelength ; and the double - peaked behavior of line profiles only occurs because the outer , slowly rotating regions do not fill in the profile at line center , as these regions are too cool to emit significantly at the wavelength of observation . while the standard steady disk temperature distribution @xmath35 r^{-3}$ ] is a power law at large radii
, it is relatively flat at distances within about twice the inner radius @xmath36 .
therefore , observations at long wavelengths which probe the outer disk where the temperature falls rapidly with radius will exhibit stronger rotational velocity variations and more double - peaked line profiles than observations at short wavelengths probing the inner , more nearly isothermal disk . in zhu et al .
( 2007 ) we were forced to use a maximum disk temperature @xmath37 k to match the sed of fu ori , which is lower than the 7200 k maximum temperature adopted in the model used by ph08 ( which was based on the earlier model by kenyon , hartmann , & hewett 1988 ) .
lowering the maximum temperature has the effect of making the flatter part of the accretion disk temperature distribution more dominant at optical wavelengths . as shown in table 1 and figure 3 , this model predicts essentially no variation of line width with lower level excitation potential and a very slight dependence on wavelength in the optical region .
( note that ph08 predict a much larger effect of line width on excitation potential than welty et al .
( 1992 ) for what should be essentially the same disk model ; the reason for the large discrepancy is unclear . ) in any case , measurement of rotation is best done through cross - correlation using suitable templates , as many of the lines used by ph08 are blends and introduce very large scatter into the model predictions ( see table 1 and figure 3 ) .
ph08 also noted that their disk model predicts very strong tio absorption bands at @xmath38 and @xmath39 which are not observed .
however , as shown in figure 2 , our lower - temperature disk model does not predict strong tio absorption bands in this region .
in addition there is evidence for @xmath39 bandhead absorption in our mike spectra , at the level predicted by our disk model .
once again this difference in the predicted disk model spectra arises simply by reducing the maximum disk temperature , which increases the importance of hot inner disk relative to the outer cool disk at the wavelength of observation .
it has long been known that many optical line profiles in fu ori are less double - peaked than predicted by simple quiescent disk models ( hartmann & kenyon 1985 ) .
there are , however , alternative possibilities to explain the profiles which do not demand abandonment of the accretion disk hypothesis .
if , as currently thought , ionized disks accrete through the action of the magnetorotational instability ( mri ; balbus & hawley 1998 ) , such disks must be turbulent .
the disk models with resolved vertical structure computed by miller & stone ( 2000 ) predict that turbulence driven by the mri in the central layers of the disk produces waves which propagate outward and shock in the upper layers .
it would be surprising if the mri did not produce significant turbulent motions in the upper atmospheric layers of the disk , which would tend to wash out the double profile structure .
hartmann , hinkle , & calvet ( 2004 ) found that that some mildly supersonic turbulence was needed to explain the @xmath40co first - overtone lines of fu ori .
it should be noted that the standard steady disk structure may not be completely applicable in the innermost disk .
standard thin - disk models predict that accretion onto a slowly rotating star should give rise to boundary - layer emission with roughly half the system luminosity ; this is not observed in fu ori ( kenyon et al . 1989 ) .
popham et al .
( 1996 ) considered disk models which suppress boundary layer radiation ; such models exhibit less doubled line profiles in inner disk regions , largely because the angular velocity of the disk departs from keplerian values near the inner disk boundary . to explain the optical spectrum of fu ori with a central star
, the star would be required to have essentially the same total system luminosity @xmath41 , and would need to have a radius roughly twice the inner radius of the disk model , @xmath42 ( zhu et al . 2007 ) .
assuming keplerian rotation for the infrared disk ,
zhu et al . estimated a central mass @xmath43 .
such a star can not be an isolated product of stellar evolution , as it has an implausibly short kelvin - helmholtz contraction time @xmath44 yr . the energy to power the star would have to come from disk accretion , which would also potentially explain the outburst ( larson 1983 ) .
however , as the ratio of optical - to - infrared rotational velocities is consistent with a keplerian profile , this implies that any central star would have to be rotating nearly at breakup ; this means that the assumption of solid - body rotation in the ph08 model is unlikely to be correct .
it is also unclear whether the accretion of a large amount of hot disk material would add enough angular momentum to spin up the outer layers of the star to breakup velocity as it expanded the outer atmosphere . in summary , the accretion disk model for fu
ori provides a coherent explanation of the observed spectral energy distribution and differential rotation over more than a decade in wavelength .
the slow rotation observed at @xmath8 m supports our previous result that the high mass accretion rate disk could extend to @xmath45 au , which is significantly larger than that predicted by the pure thermal instability theory @xcite . on the other hand
, the theory incorporating both gravitational and magnetorotational @xcite successfully predicts au scale high mass accretion rate inner disk during outbursts @xcite . with the advent of more powerful computers and sophisticated magnetohydrodynamic codes , and the assumption of mri - driven accretion
, it should be possible to explore the possibility that atmospheric turbulence and/or nonstandard inner thin disk structure can explain details of the optical line profiles .
this work is supported in part by nasa grant nnx08ai39 g and is based in part 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 science and technology facilities council ( united kingdom ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) , ministrio da cincia e tecnologia ( brazil ) and ministerio de ciencia , tecnologa e innovacin productiva ( argentina ) .
the observations were obtained with the phoenix infrared spectrograph , which was developed and is operated by the national optical astronomy observatory .
the gemini / phoenix spectra were obtained as part of program gs-2007a - c-4 .
ccccc|ccccc @xmath46 & ion & ep & hwhd & grade & @xmath46 & ion & ep & hwhd & grade + ( ) & & ( ev ) & ( km s@xmath28 ) & & ( ) & & ( ev ) & ( km s@xmath28 ) & + 5383.37 & fe i & 4.31 & -&0 & 6726.66 & fe i & 4.61 & 66.2 & 3 + 5717.83 & fe i & 4.28 & -&1 & 6767.79 & ni i & 1.83 & 65.4 & 3 + 5772.15 & si i & 5.08 & -&0 & 6810.26 & fe i & 4.61 & - & 1 + 5775.08 & fe i & 4.22 & -&1 & 6814.94 & co i & 1.96 & - & 0 + 5862.35 & fe i & 4.55 & -&0 & 6828.59 & fe i & 4.64 & 78 & 3 + 5899.29 & ti i & 1.05 & 61.3&2 & 7090.38 & fe i & 4.23 & 66.8 & 4 + 5922.11 & ti i & 1.05 & 61.8&3 & 7122.19 & ni i & 3.54 & 66.1 & 4 + 5934.66 & fe i & 3.93 & 78.4&3 & 7344.70 & ti i & 1.46 & 76.8 & 2 + 5965.83 & ti i & 1.88 & 65.9&3 & 7393.60 & ni i & 3.61 & 68 & 3 + 5987.07 & fe i & 4.80 & 71.7&3 & 7445.75 & fe i & 4.26 & -&0 + 6016.66 & fe i & 3.55 & - & 1 & 7511.02 & fe i & 4.18 & 61.7 & 4 + 6024.06 & fe i & 4.55 & - & 1 & 7525.11 & ni i & 3.63 & -&1 + 6027.05 & fe i & 4.09 & - & 0 & 7555.60 & ni i & 3.85 & 68.3&3 + 6056.01 & fe i & 4.73 & - & 1 & 7568.89 & fe i & 4.28 & 69.4&4 + 6108.11 & ni i & 1.68 & - & 1 & 7574.04 & ni i & 3.83 & -&1 + 6180.20 & fe i & 2.73 & 70 & 2 & 7586.01 & fe i & 4.31 & 65.8&5 + 6270.23 & fe i & 2.86 & 72.2 & 3 & 7727.61 & ni i & 3.68 & 65.4&4 + 6355.03 & fe i & 2.85 & 64.4 & 3 & 7780.55 & fe i & 4.47 & 65.4&4 + 6358.70 & fe i & 0.86 & 66.05 & 4 & 7788.94 & ni i & 1.95 & 64.13&4 + 6380.74 & fe i & 4.19 & 72.2 & 3 & 7855.44 & fe i & 5.06 & 62.1&4 + 6411.65 & fe i & 3.65 & 67.8 & 2 & 7912.87 & fe i & 0.86 & -&0 + 6439.08 & ca i & 2.53 & 74.5 & 3 & 7937.13 & fe i & 4.33 & 63.5&4 + 6471.66 & ca i & 2.53 & 72.8 & 3 & 8075.15 & fe i & 0.92 & -&0 + 6475.62 & fe i & 2.56 & 66.3 & 5 & 8080.55 & fe i & 3.30 & -&0 + 6569.22 & fe i & 4.73 & 67.8 & 5 & 8085.18 &
fe i & 4.45 & 64.7 & 3 + 6581.21 & fe i & 1.49 & 64.7 & 4 & 8426.51 & ti i & 0.83 & - & 0 + 6586.31 & ni i & 1.95 & -&1 & 8611.80 & fe i & 2.85 & 65.0 & 4 + 6707.89 & li
i & 0.00 & -&0 & 8621.60 & fe i & 2.95 & 63.5 & 3 + 6717.68 & ca i & 2.71 & 69.0&4 & 8648.47 & si i & 6.21 & 73.9 & 3 + 6721.85 & si i & 5.86 & 65.2&2 & + m phoenix spectrum of fu ori ( solid curve ) compared with our synthetic disk model spectrum ( upper dotted curve ) .
while the wavenumber is shown at the bottom , the wavelength in microns is shown at the top of the figure .
the lower dotted curve shows the model spectrum observed pole - on , which removes the rotational broadening , illustrating the need for rapid rotation to match the observations ( see the text ) .
the strong lines are identified in the lower model spectrum .
the arrow points to the strong co line which is missing in our model.,scaledwidth=90.0% ] | the emission of fu orionis objects in outburst has been identified as arising in rapidly accreting protoplanetary disks , based on a number of observational properties .
a fundamental test of the accretion disk scenario is that the differentially rotating disk spectrum should produce a variation of rotational velocity with the wavelength of observation , as spectra taken at longer wavelengths probe outer , more slowly rotating disk regions .
previous observations of fu ori have shown smaller rotation at near - infrared ( @xmath0 m ) wavelengths than observed at optical ( @xmath1 m ) wavelengths consistent with the assumption of keplerian rotation . here
we report a spectrum from the phoenix instrument on gemini south which shows that differential ( slower ) rotation continues to be observed out to @xmath2 m .
the observed spectrum is well matched by the prediction of our accretion disk model previously constructed to match the observed spectral energy distribution and the differential rotation at wavelengths @xmath3 m .
this kinematic result allows us to confirm our previous inference of a large outer radius ( @xmath41 au ) for the rapidly accreting region of the fu ori disk , which presents difficulties for outburst models relying purely on thermal instability .
while some optical spectra have been interpreted to pose problems for the disk interpretation of fu ori , we show that the adjustment of the maximum effective temperature of the disk model , proposed in a previous paper , greatly reduces these difficulties . |
using analysis for fundamental solutions of laplace s equation on riemannian manifolds of constant curvature , we have previously obtained antiderivatives and integral representations for associated legendre and ferrers functions of the second kind with degree and order equal to within a sign .
for instance in cohl ( 2011 ) ( * ? ? ? * theorem 1 ) , we derived using the @xmath0-dimensional hypersphere with @xmath1 an antiderivative and an integral representation for the ferrers function of the second kind with order equal to the negative degree . in cohl & kalnins ( 2012 ) ( * ? ? ? * theorem 3.1 ) , we derived using the @xmath0-dimensional hyperboloid model of hyperbolic geometry with @xmath2 , an antiderivative and an integral representation for the associated legendre function of the second kind with degree and order equal to each other . in @xcite , @xcite ,
the antiderivatives and integral representations were restricted to values of the degree and order @xmath3 such that @xmath4 is an integer . the purpose of this paper is to generalize the results presented in @xcite , @xcite for associated legendre and ferrers functions of the first and second kind , and to extend them such that the degree and order are no longer subject to the above restriction .
our extremely simple integral representations are consistent with known special values for associated legendre and ferrers functions of the first kind when the order is equal to the negative degree . throughout this paper
we rely on the following definitions .
let @xmath5 , with @xmath6 being the set of complex numbers .
if @xmath7 and @xmath8 , then @xmath9 .
the set of natural numbers is given by @xmath10 , the set @xmath11 , and @xmath12
as is the common convention for associated legendre functions ( * ? ? ?
* ( 8.1.1 ) ) , for any expression of the form @xmath13 , read this as @xmath14 for any fixed @xmath15 and @xmath16 $ ] .
let @xmath16,$ ] @xmath17 , and @xmath18 be a constant .
then we have the following antiderivative @xmath19 \nu+\frac32\end{array};\frac{1}{z^2}\right ) + c\nonumber\\[0.2 cm ] & & \hspace{2cm}=\frac{-2^{-\nu}e^{-i\pi\nu } } { \gamma(\nu+1 ) ( z^2 - 1)^{\nu/2 } } q_\nu^\nu(z)+c . \label{antiderivativecosh2r}\end{aligned}\ ] ] [ thmantifirst ] in the above expression , the gauss hypergeometric function @xmath20 can be defined in terms of the following infinite series @xmath21 c\end{array};z\right):= \sum_{n=0}^\infty \frac{(a)_n(b)_n}{(c)_n}\frac{z^n}{n!},\ ] ] ( see ( 2.1.5 ) in andrews , askey & roy 1999 ) and elsewhere on @xmath22 by analytic continuation . the pochhammer symbol ( rising factorial ) @xmath23
is defined by @xmath24 where @xmath25 .
the associated legendre function of the second kind @xmath26\to{{\mathbf c}}$ ] can be defined in terms of the gauss hypergeomertic function for @xmath27 , as ( olver _ et al .
_ ( 2010 ) ( * ? ? ?
* ( 14.3.7 ) and section 14.21 ) ) @xmath28 \nu+\frac32\end{array } ; \frac{1}{z^2 } \right ) , \label{qdefn}\ ] ] for @xmath29 and by analytic continuation of the gauss hypergeometric function elsewhere on @xmath16 $ ]
. * proof . * the antiderivative ( [ antiderivativecosh2r ] ) is verified as follows . by using @xmath30 c\end{array};z\right)=\frac{ab}{c}\,{}_2f_1 \left(\begin{array}{c}a+1,b+1\\[0.2 cm ] c+1\end{array};z\right )
\label{derhyp}\ ] ] ( see ( 15.5.1 ) in olver _
( 2010 ) @xcite ) , and the chain rule , it follows that @xmath31 \nu+\frac32\end{array } ; \frac{1}{z^2}\right ) \nonumber\\[0.2cm]&&\hspace{5.4 cm } -\frac{2(\nu+1)}{(2\nu+3)z^{2\nu+4 } } \,{}_2f_1\left ( \begin{array}{c } \nu+\frac32,\nu+2\\[0.2 cm ] \nu+\frac52\end{array } ; \frac{1}{z^2}\right ) .
\label{sum}\end{aligned}\ ] ] the second hypergeometric function on the right - hand side can be simplified using gauss relations for contiguous hypergeometric functions , namely @xmath32 c+1\end{array};z\right)=\frac{c}{a - b}\left[{}_2f_1\left(\begin{array}{c}a , b+1\\[0.2 cm ] c\end{array};z\right ) -{}_2f_1\left(\begin{array}{c}a+1,b\\[0.2 cm ] c\end{array};z\right)\right ] \label{sechypcontig}\ ] ] ( see p. 58 in erdlyi _
et al . _ ( 1981 ) @xcite ) , and @xmath33 c\end{array};z\right)=\frac{b - a}{b}\,{}_2f_1\left(\begin{array}{c}a , b \\[0.2cm]c\end{array};z\right ) + \frac{a}{b}\,{}_2f_1\left(\begin{array}{c}a+1,b\\[0.2cm]c\end{array};z\right ) \label{thirdhypcontig}\ ] ] ( see ( 15.5.12 ) in olver _
( 2010 ) @xcite ) . after simplification and utilization of @xmath34 b\end{array};z\right)=(1-z)^{-a}\ ] ] ( see ( 15.4.6 ) in olver _
( 2010 ) @xcite ) , the right - hand side of ( [ sum ] ) simplifies to @xmath35 .
the equality of the gauss hypergeometric function in the antiderivative ( [ antiderivativecosh2r ] ) , in terms of the associated legendre function @xmath36 follows directly from ( [ qdefn ] ) .
@xmath37 a straightforward consequence of the antiderivative ( [ antiderivativecosh2r ] ) is the following integral representation for the associated legendre function with degree and order equal , namely @xmath38 where @xmath39 . using the negative order relation for associated legendre functions of the second kind ( *
* ( 14.9.14 ) ) @xmath40 we derive @xmath41 using the whipple relation for associated legendre functions ( cf .
* ( 14.9.16 ) ) ) @xmath42 where the associated legendre function of the first kind @xmath43\to{{\mathbf c}}$ ] is defined in ( * ? ? ?
* ( 14.3.6 ) ) , we obtain @xmath44 the above expression is equivalent to @xmath45 & & \hspace{3.2cm}+\frac{2^{\nu+1}\gamma(\nu+1)(z^2 - 1)^{\nu/2 } } { \pi}\sin(\nu\pi ) \int_{z}^\infty \frac{dw}{(w^2 - 1)^{\nu+1 } } , \label{pnunuz}\end{aligned}\ ] ] where @xmath39
. an interesting definite integral follows from the behavior of the above integral representation near the singularity at @xmath46 , namely @xmath47 for @xmath48 . using the negative order relation for associated legendre functions of the first kind ( cf .
* ( 14.9.12 ) ) ) , one has @xmath49 . \label{pnumnu}\ ] ] after replacement of ( [ qnunuz ] ) and ( [ pnunuz ] ) in ( [ pnumnu ] ) we obtain @xmath50,\ ] ] which reduces to the special value ( * ? ? ?
* ( 14.5.19 ) ) @xmath51
let @xmath52 @xmath53 and @xmath18 be a constant
. then we have the following antiderivative @xmath54 \frac32\end{array};{x^2}\right ) + c = \frac{2^\nu\gamma(\nu+\tfrac12)}{\sqrt{\pi}\,(1-x^2)^{\nu/2 } } { \rm q}_\nu^{-\nu}(x)+c .\ ] ] [ antiderivthmsecond ] in the above expression , the ferrers function of the second kind ( associated legendre function of the second kind on - the - cut ) @xmath55 is defined in ( * ? ? ?
* ( 14.3.12 ) ) , and for @xmath56 we have @xmath57 \frac32 \end{array};x^2 \right ) .
\label{ferrerssecond}\ ] ] * proof . * the gauss hypergeometric function in the antiderivative follows using ( [ derhyp ] ) , ( [ sechypcontig ] ) , ( [ thirdhypcontig ] ) , as in the proof of theorem [ thmantifirst ] , with the ferrers function following directly using ( [ ferrerssecond ] ) .
@xmath37 the following integral representation for the ferrers function of the second kind is an obvious consequence of theorem [ antiderivthmsecond ] , namely @xmath58 a definite - integral result near the singularity at @xmath59 follows using ( [ ferrerssecond ] ) , ( [ intrepferrers2numnu ] ) , and gauss s sum ( * ? ? ?
* ( 15.4.20 ) ) @xmath60 for @xmath61 , namely @xmath62 for @xmath63 . the well - known special value ( see ( * ? ?
* ( 14.5.18 ) ) ) @xmath64 where @xmath65 is the ferrers function of the first kind ( associated legendre function of the first kind on - the - cut ) defined in ( * ? ? ?
* ( 14.3.1 ) ) , in conjunction with ( * ? ? ?
* ( 8.737.1 ) ) , yields the following integral representation @xmath66.\ ] ] finally using the negative order relation for the ferrers functions of the second kind @xcite , namely @xmath67,\ ] ] we have the following integral representation of the ferrers function of the second kind with degree and order equal @xmath68 & & \hspace{5.0cm}+2^\nu\gamma(\nu+1)\cos(\nu\pi)(1-x^2)^{\nu/2 } \int_0^x\frac{dw}{(1-w^2)^{\nu+1}}.\end{aligned}\ ] ]
part of this work was conducted while h. s. cohl was a national research council research postdoctoral associate in the applied and computational mathematics division at the national institute of standards and technology , gaithersburg , maryland , u.s.a . | we obtain antiderivatives and complex integral representations for associated legendre functions and ferrers functions ( associated legendre functions on - the - cut ) of the first and second kind with degree and order equal to within a sign .
associated legendre functions ; ferrers functions ; integral representations ; gauss hypergeometric function 30e20 ; 35c05 ; 26a36 |
quantum control is aimed at designing external pulses in order to achieve efficient transfers between the states of the quantum system under study @xcite .
this task is crucial in atomic and molecular physics , and has many applications extending from photochemistry to quantum computation .
quantum control has attracted attention among the physics and chemistry communities @xcite , but also in applied mathematics for the development of new theoretical methods . in this context ,
optimal control theory ( oct ) can be viewed as the most accomplished way of designing control fields @xcite .
several modifications of the standard optimal control algorithms have been brought forward to account for experimental constraints @xcite , such as the non linear interaction of the system with the field @xcite , the question of spectral constraints @xcite and the robustness with respect to one or several model parameters @xcite .
recently , we have shown how optimal control strategies can be extended to enforce the constraint of time - integrated zero - area on the control field @xcite .
this constraint is a fundamental requirement in laser physics , as shown in different experimental and theoretical studies @xcite .
basically this effect can be related to the fact that the dc component of the control field is not a solution of maxwell s equation .
we refer the reader to the first part of the paper for a complete discussion .
note that this point is particularly crucial in the thz regime , or with laser pulses accommodating only few optical cycles @xcite .
the corresponding laser sources are by now commonly used in quantum control @xcite . up to now
, the majority of the theoretical papers on quantum control does not consider this zero - area requirement @xcite , which may lead to non - physical electromagnetic control fields and is problematic in view of experimental implementations .
in addition , imposing such a constraint to the control scheme may force the optimization algorithm to reach more efficient external fields achieving better transfer .
these arguments show the importance of the methods and of the results presented in this work , in particular to fill the existing gap between theory and experiment .
our preliminary study on the subject @xcite is a methodology - oriented paper , focusing mainly on the technical aspects of the optimization algorithms ( local and optimal approaches ) as briefly illustrated on two typical molecular processes ( orientation and dissociation ) .
the present paper thoroughly expands this initial work @xcite by providing an extended numerical investigation and a detailed physical analysis of the dynamics of two specific molecular systems .
the article is organized as follows .
the physical origin of the time - integrated zero - area constraint is presented in sec.[sec2 ] .
for completeness the principles of the optimization algorithms are also briefly outlined , full technical details being referred to ref.@xcite .
section [ sec3 ] focuses on the control of molecular orientation , with the co molecule as an illustrative example .
section [ sec4 ] is devoted to the dissociation of heh@xmath0 involving the control of a given fragmentation channel .
conclusions and prospective views are given in a final section [ sec5 ] .
the zero - area constraint for laser pulses although well - known , is rarely given a thorough and clear argumentation . for completeness and pedagogical purposes ,
this section is devoted to the presentation of such a proof .
the body of the proof is made of two parts : the calculation of the time integrated electromagnetic field and the physical interpretation of the result .
we refer to the spatio - temporal electric field amplitude @xmath1 and its fourier transform @xmath2 , where the corresponding fourier conjugate variables are time @xmath3 and frequency @xmath4 , on the one hand , and space @xmath5 and momentum @xmath6 vectors , on the other hand .
these quantities are related through : @xmath7 together with the usual relation between the frequency and the wave vector : @xmath8 using these notations , the time integrated field area is given by : @xmath9 which , upon the inversion of the integration order , leads to : @xmath10 the last summation on the right - hand - side is nothing but the dirac function : @xmath11 equation ( [ a.4 ] ) can then be simplified after integration over time and frequency as : @xmath12 it is clear from eq.([a.2 ] ) that a null frequency @xmath13 has as a consequence a null momentum ( @xmath14 ) .
we finally obtain @xmath15 the physical interpretation of eq .
( [ a.7 ] ) is as follows .
the term @xmath16 actually represents a non - oscillating ( @xmath17 ) , non - propagating ( @xmath18 ) dc stark field . assuming a non - zero value for such a field requires , from the corresponding maxwell equation , the necessary existence of a spatial electronic charge distribution .
in particular , finite charges placed at finite distances may create the dc stark field in consideration .
one may also invoke a charge distribution set at infinite distance , as limiting asymptotic conditions .
but , of course , within this hypothesis , a non - zero stark field would necessitate an overall distributed infinite electric charge .
ultimately , in systems where electric charges are not introduced on purpose , a dc stark field can not be created , and @xmath19 the goal of this section is to show how zero - area control fields can be designed .
the three proposed methods are based on the optimization of the parameters of a functional form associated with the control problem .
we consider a quantum system interacting with an electromagnetic field whose dynamics is described by the following time - dependent schrdinger equation : @xmath20 where @xmath21 and @xmath22 are the field - free hamiltonian and the interaction term , respectively , and @xmath23 the control field .
the units used throughout this paper are atomic units .
let @xmath24 be the initial state and @xmath25 the total duration of the control .
the goal of the control problem is to maximize the expectation value @xmath26 of a given observable @xmath27 at time @xmath28 .
the first approach consists in introducing a closed - form expression for the control field depending on a finite number of parameters denoted @xmath29 : @xmath30 with @xmath31 .
the ensemble @xmath32 is chosen such that the constraint on the time - integrated area , @xmath33 , is satisfied .
the optimal values of the parameters @xmath29 are determined in a second step by using gradient @xcite or global optimization procedures such as genetic algorithms @xcite .
this approach , which can be very efficient in some cases , is however highly dependent on the parametrization used to describe the control field .
a more general method is based on an extension of the optimal and local optimization algorithms , which enforces the zero - area constraint through the introduction of a lagrange multiplier .
such algorithms are proposed and investigated in ref .
for the sake of completeness of the paper , we briefly outline below the principles of the different optimization procedures and we refer the interested reader to our preceding work for technical details on the numerical algorithms @xcite . the optimal control problem is defined through a cost functional @xmath34 : @xmath35 ^ 2.\ ] ] which allows us to maximize the expectation value @xmath26 at time @xmath28 of a given observable @xmath27 , while penalizing the total energy of the control field and enforcing the zero - area constraint .
the novelty of the computational scheme resides in the introduction of a new lagrangian multiplier @xmath36 to account for the zero - area constraint . in eq .
( [ cost ] ) , the positive parameters @xmath36 and @xmath37 , expressed in a.u . ,
weight the different parts of @xmath38 with respect to the expectation value of @xmath27 , and penalize the area of the field ( @xmath36- term ) and its energy ( @xmath37- term ) .
larger is @xmath36 and closer is the time - area of the control field to zero . in eq .
( [ cost ] ) , @xmath39 is a reference pulse and @xmath40 an envelope shape , which can be chosen as @xmath41 . as usual , note that the function @xmath40 is introduced in the cost functional in order to ensure the smooth switch on and off of the field at the beginning and at end of the control . starting from this new cost functional , it is straightforward to derive a standard iterative algorithm based on krotov ( as used in this work ) @xcite or gradient procedure @xcite .
the basic ingredient of the optimization procedure is the definition , at each step , of a new control field . at step @xmath42 ,
we get : @xmath43}{2\lambda}-\frac{\mu}{\lambda}s(t)a_k , \label{eq : newfield}\ ] ] where @xmath44 is the state of the system at the @xmath42th iteration , @xmath45 the adjoint state obtained from backward propagation of the target @xmath46 taken as an initial state for eq .
( [ eqschrodinger ] ) , @xmath47 and @xmath48 the control fields at steps @xmath42 and @xmath49 , respectively , and @xmath50 the corresponding time - integrated area . only the last term of the right - hand side of eq .
( [ eq : newfield ] ) is different from a standard procedure .
the local control theory ( lct ) @xcite can also be extended along the same lines by considering the following lyapunov function which accounts for the zero - area constraint : @xmath51 where @xmath27 is any operator such that @xmath52 .
\label{eq : comuto}\ ] ] to ensure the monotonic increase of @xmath53 at any time @xmath3 , i.e. @xmath54 , the control field is defined as follows : @xmath55|\psi(t)\rangle-2\mu a(t)\big),\ ] ] where @xmath56 is a positive parameter used to limit the amplitude of @xmath23 .
we will also use in the following the parameter @xmath57 .
finally , note that the two proposed optimization procedures can only reduce the time - integrated area without completely removing it .
a filtering process can further be used to accurately design a zero - area field .
this section is devoted to the control of the orientation dynamics of polar diatomic molecules @xcite with the constraint of zero - area fields .
the two different approaches introduced in sec .
[ sec2 ] will be considered and discussed .
we introduce a family of pulses characterized by a closed - loop expression depending on two parameters , which can be adjusted to enhance the degree of orientation .
a second option is based on the optimization algorithms of sec .
[ sec2 ] which enforce the zero - area constraint through a lagrange multiplier .
no analytic expression of the optimal field can be derived in this case .
we consider a molecule described in a rigid - rotor approximation interacting with @xmath23 , a linearly polarized electromagnetic pulse along the @xmath58- axis of the laboratory frame .
the electric field is assumed to be in the thz regime .
the co molecule in its ground vibronic state is taken as an illustrative example . at zero temperature , the dynamics
is governed by the time - dependent schrdinger equation ( [ eqschrodinger ] ) where , in a linear approximation , @xmath59 and @xmath60 .
the parameter @xmath61 is the rotational constant of the molecule , @xmath62 the angular momentum operator and @xmath63 the molecular permanent dipole moment .
the spatial position of the diatomic molecule is given in the laboratory frame by its spherical coordinates @xmath64 , @xmath65 being the angle between the molecular axis and the polarization vector , and @xmath66 the corresponding azimuthal angle . the hilbert space associated with the dynamical system is spanned by the spherical harmonics @xmath67 , with @xmath68 and @xmath69 .
we also recall that , due to the cylindrical symmetry of the problem , the projection @xmath70 of the total angular momentum on the field polarization axis is a good quantum number , so that @xmath71 does not depend on the angle @xmath66 . when the laser is switched on at @xmath72 , the initial condition is given by @xmath73 .
the expectation value @xmath74 is usually taken as a quantitative measure of orientation . in the non - zero temperature case ,
the time evolution of the molecular system is described by the density operator @xmath75 solution of the von neumann equation : @xmath76.\ ] ] the initial condition is a boltzmann distribution at temperature @xmath77 and the degree of orientation is given by the expectation value expressed in terms of @xmath75 : @xmath78 $ ] .
numerical values of the molecular parameters are taken as @xmath79 @xmath80 and @xmath81 a.u .
the rotational period , @xmath82 , is of the order of 8.64 ps or @xmath83 a.u . in the numerical computations ,
the maximum number of @xmath84- levels populated is of the order of 12 .
we first consider a class of pulses characterized by a closed - from expression in order to maximize the molecular orientation of the co molecule at zero temperature .
the electric field of this family can be written as follows @xcite : @xmath85\\ & & e(t)=0,~t\notin [ -\delta/2,\delta/2 ] , \nonumber\end{aligned}\ ] ] with @xmath86 the pulse amplitude , @xmath87 its duration and @xmath88 its carrier wave frequency .
it is straightforward to check that for any values of @xmath87 and @xmath88 , @xmath23 has actually a zero time - integrated area .
the choice of this form is motivated by the experimental feasibility of such pulses in the thz regime @xcite .
we start our analysis by a general study of the degree of orientation that can be achieved within this class of fields . the field amplitude is fixed and corresponds to a peak intensity of 20 tw/@xmath89 , while the parameters @xmath87 and @xmath88 belong to the intervals @xmath90 $ ] and @xmath91 $ ] thz , respectively . the maximum orientation achieved during the field - free evolution
is indicated in fig . 1 by a circle .
we observe that the global degree of orientation is generally low , except for a zone of high orientation around @xmath92 thz and @xmath93 .
the maximum of @xmath94 obtained is of the order of @xmath95 for @xmath96 thz and @xmath97 .
the size of the high orientation region shows the robustness of the control pulse with respect to experimental imperfections while setting the parameters @xmath87 and @xmath88 .
( vertical color code ) for the co molecule as a function of the parameters @xmath88 and @xmath87 defined in eq.([family ] ) .
the circle indicates the pulse for which the maximum degree of orientation is achieved . ] in a second step of the investigation , we use the best control field derived from the results of fig .
[ fig1 ] as a guess field for the optimal control algorithm .
to be applied , this algorithm requires the definition of a target state . here
, we introduce the target operator @xmath77 defined as @xmath98 where @xmath99 is taken as the delay between the end of the guess pulse and the time where @xmath100 reaches its first maximum during the field - free evolution of the system .
figure [ fig : evol_cos_i ] displays the time evolution of @xmath101 and gives the definition of the time @xmath99 .
this figure also illustrates how @xmath99 is chosen
. induced by the guess field .
the time @xmath99 is given by the relation @xmath102 . ] in numerical calculations , the intensity of the guess field and the parameter @xmath37 are fixed to 20 tw/@xmath103 and 1 , respectively , for both optimizations , with and without zero - area constraint .
the delay @xmath99 in eq .
( [ eq : tar_op ] ) is set to @xmath104 .
the dynamics under the optimized fields is shown in fig . [
fig : jt_area](a ) , comparing the effect of standard and zero - area constraint algorithms . for @xmath105 ( optimization without zero - area - solid thin blue line ) and for @xmath106 ( solid thick black line - @xmath107 a.u . ) .
the red dashed line represents the orientation dynamics induced by the initial guess field .
the panels ( b ) and ( c ) display the corresponding control fields and their fourier transform . ]
figure [ fig : jt_area](b ) shows the guess and the optimized fields with and without the zero - area constraint .
note that the global shape of the two optimized fields is similar to the guess field , except for a small oscillatory behavior .
the optimized pulse without the zero - area constraint leads to higher orientation ( @xmath108 ) than the pulse with the zero - area constraint ( @xmath109 ) , but the time - integrated area is divided by a factor 60 in the second case , going from -20.57 a.u .
to -0.36 a.u . advocating for experimental feasibility .
the very good orientation achieved demonstrates the efficiency of the optimal control algorithms , even if the area of the field is no more strictly zero .
as expected , we observe in fig .
[ fig : jt_area](c ) that the fourier transform of the three fields is equal or nearly equal to 0 at @xmath110 .
since the optimized fields are very close to the guess pulse , their fourier transforms show similar features . however , the optimization slightly shifts the fourier transform towards the high frequencies .
the efficiency of the zero - area constraint algorithm is also checked for the co molecule at non zero temperature .
this is a much more difficult task than controlling the orientation at zero temperature .
we discuss only the case of a long optimization time , @xmath111 .
we have considered as initial field a closed - form expression with zero area , which can be written as the sum of three hermite polynomials ( see fig .
[ fig : jt_area_t30k](b ) for a representation of this pulse ) .
we have changed the maximum intensity from 20 tw/@xmath89 to 2 tw/@xmath89 .
the parameters @xmath37 and @xmath36 are set to 20 a.u . and
@xmath112 a.u .
respectively . the temperature is fixed to 30 k. figure [ fig : jt_area_t30k](a ) displays the time evolution of the expectation value of @xmath113 induced by the guess and optimized fields .
the dynamics under the optimized fields have similar features , but are very distinct from the one induced by the guess field .
but for the non - zero temperature case .
] figure [ fig : jt_area_t30k](b ) compares the corresponding optimized fields together with the guess field .
as could be expected , the optimized fields show similar features . in this example , the area of the optimized field with the zero - area constraint is two orders of magnitude smaller than the one obtained without the zero - area constraint .
this is a remarkable result since the time - integrated area is largely reduced while preserving a satisfactory orientation of the order of 0.2 .
the price to pay for increasing the final degree of orientation can be seen in the fourier transform of the optimized pulses , which have a much more complicated structure with an oscillatory behavior at low frequency .
these additional low frequencies found by the algorithm correspond to the slow oscillations of the optimized fields which appear after @xmath114 . by filtering out such oscillations , we have checked that this oscillatory behavior is essential to produce a high degree of orientation .
following ref .
@xcite , we observe that the low frequency distribution of the optimized field coincides with the rotational resonance frequencies .
this suggests an interpretation for the origin of the oscillatory behavior and a possible control mechanism based on the excitation of these different frequencies .
another important application of local and optimal control strategies is illustrated on molecular photodissociation .
due to the short duration of the pulses ( as compared with the rotational period ) a frozen rotation approximation is valid .
in addition , the molecule is assumed to be pre - aligned along the @xmath58-direction of the laboratory frame .
therefore , the diatomic system is described by its reduced mass @xmath70 and the internuclear distance @xmath115 .
we aim at controlling the photodissociation of heh@xmath0 through the singlet @xmath116 excited states leading to he@xmath117 fragment in the @xmath118 shell .
we shall consider only parallel transitions among the singlet @xmath116 states induced by the @xmath119 dipole operator assuming the internuclear axis pointing along the @xmath58-direction .
the adiabatic potential energy curves , the radial nonadiabatic couplings @xmath120 and the adiabatic transition elements @xmath121 of the @xmath119 dipole operator have been computed in ref .
[ fig : heh_pot ] ( a ) displays the adiabatic potential energy curves .
the partial photodissociation cross sections computed in ref .
@xcite are displayed in fig .
[ fig : heh_pot ] ( b ) .
states of heh@xmath0 leading to fragments in the @xmath118 shell .
the target states are : h@xmath0 + he(1s2s ) ( green dashed line ) , h@xmath0 + he(1s2p ) ( red long dashes dots ) .
( b ) partial photodissociation cross sections together with an enlargement given as inset .
the legend is the same as in panel ( a ) .
] dynamics is performed in the diabatic representation obtained from the adiabatic - to - diabatic transformation matrix @xmath122 which has been derived by integrating @xmath123 from the asymptotic region where both representations coincide .
the total hamiltonian introduced in eq .
( [ eqschrodinger ] ) involves here @xmath124 and @xmath125 where @xmath126 are the potential matrix elements as obtained by diabatization of the adiabatic potential energies using the transformation matrix @xmath122 .
the parameter @xmath127 is the number of diabatic electronic states @xmath128 under consideration and @xmath129 .
@xmath130 are the diabatized dipole transition matrix elements .
the initial state @xmath131 is the lowest vibrational state of the ground electronic adiabatic state .
the goal is to enhance the yield in he@xmath117 in the @xmath132= 2 shell through the dissociation channels leading to he@xmath133 , or to he@xmath134 @xcite .
these two target asymptotic states are denoted by @xmath135 ( fig .
[ fig : heh_pot ] , green dashed curve ) and @xmath136 ( fig .
[ fig : heh_pot ] , red long dashes dots ) . in a first attempt we use a zero area gaussian pulse with a carrier frequency chosen from the photodissociation cross section maximum .
we limit the total integrated intensity for further comparison with the oct strategy .
the yields remain very weak , of the order of 3 % only , close to the value predicted by the fragmentation cross section .
we then examine the efficiency of the zero - area constraint in both local and optimal control approaches .
the local control can be considered as a very interesting first step before using oct . in the presence of nonadiabatic interactions , the operator @xmath27 referred to in eq .
( [ eq : jlc ] ) requires to be chosen carefully since it has to commute with the field - free hamiltonian @xmath21 ( see eq .
( [ eq : comuto ] ) ) .
the projectors on either adiabatic or diabatic states are thus not appropriate in lct since they do not commute with this hamiltonian due to kinetic or potential couplings , respectively @xcite .
this crucial problem can be overcome by using projectors on eigenstates of @xmath21 , i.e. on scattering states correlating with the controlled exit channels . in this example , the operator @xmath27 takes the form : @xmath137 where @xmath138 represents the two channels leading to the target he@xmath139 fragments , the objective being @xmath140 .
the local control field now reads @xmath141 -2 \tilde{\mu } a(t),\ ] ] involving two adjustable parameters @xmath56 and @xmath142 .
the ingoing scattering states @xmath143 are estimated using a time - dependent approach based on mller - operators @xcite defined by @xmath144 where @xmath145 is the outgoing plane wave in channel @xmath146 with energy @xmath147 and @xmath148 is the hamiltonian operator of the fragments where all couplings have vanished asymptotically .
this control strategy remains local in time but can preemptively account for nonadiabatic transitions that occur later in the dynamics .
the photodissociation cross section @xcite shows that there is no spectral range where the he@xmath139 dissociation channels dominate ( see fig . [
fig : heh_pot](b ) ) .
the local control without any constraint ( @xmath149 ) finds a very complicated electric field which begins by a regular oscillatory pattern followed after 10 fs by a complex , positive real component shape whose area is obviously not zero ( see blue thick curve in fig . [
fig : heh_field](a ) ) .
this erratic positive structure found in the lct field can be interpreted as a stark field .
we therefore choose this example to check the efficiency of the zero - area constraint algorithm .
we first use the zero - area algorithm with @xmath150 a.u . and different values of @xmath142 (
some examples are given in ref .
figure [ fig : heh_field](a ) ( green thin curve ) shows the pulse for @xmath142 = 0.05 a.u .
the algorithm efficiently reduces the stark structure without completely removing it .
the average objective ( eq . ( [ newproj ] ) ) for the two cases without ( blue thick curve ) and with the area constraint ( green thin curve ) are displayed in fig .
[ fig : heh_field](d ) .
the objective is divided by about 2/3 . as shown in ref .
@xcite , increasing @xmath142 to still reduce the stark component decreases the objective so that a compromise has to be found .
a complementary brute force strategy consists in removing the main part of the stark component by filtration of near - zero frequencies .
starting from the initial lct pulse , this already provides a large correction to the non vanishing area . the field after filtration of the low frequencies
is shown in fig .
[ fig : heh_field](b ) ( red thin curve ) . to estimate the efficiency of the filtered pulse
, we show the occupation of the two target adiabatic channels during the propagation in fig .
[ fig : heh_field](e ) ( red thin curve ) .
the final value of 3.75@xmath151 is notably lower than the asymptotic value of the local objective 8.55 @xmath151 ( blue thick curve in fig .
[ fig : heh_field](d ) ) .
figure [ fig : heh_field](b ) ( black thick curve ) shows the pulse with @xmath142 = 0.05 a.u .
after a subsequent filtering of the low - frequency components ( compare with the green thin curve in fig . [
fig : heh_field](a ) ) .
the resulting regular profile confirms the efficiency of this mixed strategy .
the population in the selected adiabatic channels with this filtered pulse is the black thick curve in fig .
[ fig : heh_field](e ) .
the price to pay for reducing the pulse area is always a decrease of the target yield , but the best compromise is obtained when using both area constraint algorithm and residual filtering .
we first use the zero - area algorithm with @xmath150 a.u . and
different values of @xmath142 ( some examples are given in ref.@xcite ) .
figure [ fig : heh_field](a ) ( green curve ) shows the pulse for @xmath142 = 0.05 a.u .
the algorithm efficiently reduces the stark structure without completely removing it .
the average objective ( eq . ( [ newproj ] ) ) for the two cases without ( blue thick curve ) and with area constraint ( green thin curve ) are drawn in fig .
[ fig : heh_field](d ) ) .
the objective is divided by about 2/3 . as shown in ref.@xcite ) , increasing @xmath142 to still reduce the stark component decreases the objective so that a compromise has to be found .
a complementary brute force strategy consists in removing the main part of the stark component by filtration of near - zero frequencies .
starting from the initial lct pulse , this already provides a large correction to the non vanishing area .
the field after filtration of the low frequencies is shown in fig .
[ fig : heh_field](b ) ( red thin curve ) . to estimate the efficiency of the filtered pulse
, we show the occupation of the two target adiabatic channels during the propagation in fig .
[ fig : heh_field ] ( e ) ( red thin curve ) .
the final value of 3.75@xmath151 is notably lower than the asymptotic value of the local objective 8.55 @xmath151 ( blue curve in fig .
[ fig : heh_field ] ( d ) ) .
figure [ fig : heh_field](b ) ( black thick curve ) shows the pulse with @xmath142 = 0.05 a.u .
after a subsequent filtering of the low - frequency components ( compare with the green curve in figure [ fig : heh_field](a ) ) .
the resulting regular profile confirms the efficiency of the mixed method .
the population in the selected adiabatic channels with this filtered pulse is the thick black curve in fig .
[ fig : heh_field ] ( e ) .
the price to pay for reducing the pulse area is always a decrease of the target yield , but the best compromise is obtained when using both area constraint algorithm and residual filtering . , green thin line : with zero - area constraint @xmath152 a.u . , ( b ) lct with residual filtering , red thin curve : @xmath153 a.u . ,
black thick curve : @xmath154 a.u .. ( c ) oct starting from the local field after filtration ( red thin line of panel ( b ) ) , red thin line : @xmath155 a.u .
lower panels : evolution of the objectives during the control for the different strategies .
( d ) the lct objective is the population in the selected scattering states , blue thick line : without constraint @xmath153 , green thin line : with zero - area constraint @xmath152 a.u . , ( e ) population in the adiabatic target states during the propagation with the filtered lct pulses , red thin line : @xmath153 a.u . , black thick line : @xmath154 a.u .
( f ) population in the target states during the propagation with the oct pulse , red thin curve : diabatic representation , blue thick line : adiabatic representation .
the asymptotic values give the he@xmath139 yields . ] in a second step , we explore the oct strategy .
note that this procedure only refers to the zero - area algorithm without any subsequent filtration .
the yield obtained with guess gaussian fields increases only weakly while better results are obtained when the trial field is the lct pulse .
we choose the lct pulse after filtration ( red thin curve in fig .
[ fig : heh_field](b ) ) as guess field .
note that the oct strategy only uses the zero - area algoritm without any subsequent filtration .
the lct field can be computed as long as the components in the excited states have some amplitude in the range covered by the initial ground vibrational state ( roughly speaking , the franck condon region ) .
this leads to a field vanishing after about 20 fs .
in the global oct strategy , the field is optimized on a time which may be longer .
this opens the flexibility to exploit additional transitions towards the target states .
we choose a final time @xmath156 fs .
the spatial grid is calibrated so that the target wave packet components do not reach the absorbing potential in the asymptotic region .
the oct objective is simply built here from the projector onto the diabatic states and the operator @xmath27 takes the form : @xmath157 as the objective is defined with the wave packet at the final time @xmath158 , this corresponds to the required optimization of the decoupled scattering states . at each iteration
, the final condition of the lagrange multiplier is the asymptotic components in the target channels and no amplitude in all the other ones .
the parameter @xmath37 is chosen automatically by constraining the integrated intensity to 0.06 a.u .
( see ref .
the field corresponding to the best @xmath155 a.u .
is shown in fig .
[ fig : heh_field](c ) with a yield reaching @xmath159 . as the simulation is performed in the diabatic basis set , the time evolution of the objective ( see red thin curve in fig . [
fig : heh_field](f ) ) corresponds to the population in the sum of these two diabatic channels .
the very strong oscillations reveal that the mechanism found by oct in the last step is strongly non diabatic because the selected states are coupled with the other states during the process and decouple only at the end of the control .
the mechanism is more simple in the adiabatic representation as can be seen by the evolution of the total population in the two adiabatic states correlated to the target fragments ( blue thick curve in fig .
[ fig : heh_field](f ) ) . fig .
[ fig : heh_pop ] compares the occupation of the adiabatic electronic states during the propagation with the guess field ( panel ( a ) ) and the best zero - area criterion optimal control pulse ( panel ( b ) ) .
the increase of the global target in oct mainly comes from the enhancement of the @xmath135 ( he*@xmath160 ) ( green dashed curve ) .
oct also reduces the unwanted transitions towards all other channels .
the lower panels in fig .
[ fig : heh_pop ] show the spectrograms of the filtered lct and oct fields .
the main operating frequency of the local field corresponds to that predicted by the photodissociation cross section ( at about 1.3 a.u . ) for maximizing both channels .
the inset in fig .
[ fig : heh_pot](b ) shows that this frequency corresponds to the maximum yield for the fragment @xmath135 ( he*@xmath161 ) . the additional mechanism due to a non optical stark effect
is thus suppressed by filtering very low frequencies .
the oct field first uses a low frequency component centered at about 0.8 a.u .
this frequency favors of channel @xmath135 ( he*@xmath160 ) which explains the steep increase of that population and the vanishing influence of channel @xmath136 ( he*@xmath161 ) .
the @xmath162 channel is also more involved .
after 20 fs , when the wave packet is out of the franck condon region , one observes a new mechanism proceeding via transitions between the target and the @xmath163 and @xmath136 channels .
these transitions require lower frequencies ( about 0.3 a.u . ) corresponding to the gap between the states ( see fig .
[ fig : heh_pot](a ) ) . .
the legend is the same for panels ( a ) and ( b ) .
the target states are : h@xmath0 + he(1s2s ) ( green dahed line ) , h@xmath0 + he(1s2p ) ( red dashes dots ) .
lower panels : spectrograms of the fields .
a color code with an arbitrary unit is given in each panel ( c ) or ( d ) to estimate the relative intensities .
( a ) local control after filtration taken as guess field for oct ( red thin curve in fig .
[ fig : heh_field ] ( b ) ) ( b ) optimal control with @xmath155 a.u .
( red thin curve in fig .
[ fig : heh_field ] ( c ) ) ]
after having discussed the physical origin of the time - integrated zero - area constraint on the laser control of molecular dynamics , we show that this fundamental requirement can be included in the standard optimization computational schemes .
a detailed description of the dynamics achieved with such zero - area control fields is given and applied to two specific examples of molecular dynamics , namely the control of molecular orientation and that of molecular fragmentation .
very encouraging results have been obtained even in the case of complicated quantum systems . in particular ,
we have derived for molecular orientation a closed - form expression of the control field depending only on two free parameters .
the zero - area constraint is satisfied for any value of these parameters . at zero temperature
, this approach reveals to be very efficient even when compared with the optimal solution .
however , we have observed that the modified optimal control algorithm used in this work remains the best tool to handle more involved control problems , which can not be solved by lct or analytical fields with a sufficient efficiency .
this work and the possibility of including experimental constraints in optimal control algorithms pave the way for future experimental implementations in quantum control .
in other words , such results help in bridging the gap between control theory and control experiments .
+ + * acknowledgments * s. v. acknowledges financial support from the fonds de la recherche scientifique ( fnrs ) of belgium .
financial support from the conseil rgional de bourgogne and the quaint coordination action ( ec fet - open ) is gratefully acknowledged by d. s. and m. n .. we thank the cost xlic action .
o. a. acknowledges support from the european union ( project no .
itn-2010 - 264951 , corinf ) . | the constraint of time - integrated zero - area on the laser field is a fundamental , both theoretical and experimental requirement in the control of molecular dynamics . by using techniques of local and optimal control theory ,
we show how to enforce this constraint on two benchmark control problems , namely molecular orientation and photofragmentation . the origin and
the physical implications on the dynamics of this zero - area control field are discussed . |
when a quantum system undergoes cyclic evolution motion in parameter space , its wave function acquires a geometric phase which strongly influences the transport property of the system .
the best known example of such phase is the aharonov - bohm ( ab ) phase which is the relative phase shift between two charged particle paths enclosing a magnetic flux . in the last few years , the aharonov - casher ( ac )
phase,@xcite another example of a geometric phase , has been studied both experimentally @xcite and theoretically @xcite in spin - orbit ( so ) coupled mesoscopic semiconducting rings . as the charge - spin dual of the ab effect ,
the ac phase in semiconductors has attracted great attention because it can be easily controlled by a gate voltage .
so far , most of the literature has focused on rashba so coupled rings attached to normal metal leads .
the ac phase has been confirmed in these type of systems by observing the conductance oscillation as a function of the gate voltage that is related to the so interaction ( soi ) strength .
however , due to the large background current , the typical amplitude of the oscillation is no more than 10% of the total observed conductance.@xcite the ac phase has also been discussed for magnetic vortices in type - ii superconductors and josephson junction array.@xcite recent studies in the effect of soi on josephson current@xcite has focused on heterostructures comprised of a two - dimensional electron gas ( 2deg ) or a quantum dot placed between two superconducting leads .
a natural and interesting step is to combine the ac and josephson effects to explore new regimes in semiconductor - superconductor structures , which has not been explored to our knowledge . in this article
, we study the ac effect in a josephson current traversing a ring with rashba type so interaction .
the josephson current in a superconducting - normal - superconducting ( s / n / s ) junction mainly originates from the andreev reflection,@xcite ( electrons being retroreflected to holes ) in two superconducing / normal ( s / n ) interfaces .
the excitation spectrum below the superconducting gap consists of discrete energy levels , called andreev levels.@xcite andreev levels are affected by both the phase and the transmission coefficient of electrons and holes in the normal region.@xcite we study here the effects of soi on the andreev levels and the corresponding josephson current through the ring , illustrated in fig
. [ ring ] , both analytically and numerically .
our calculations predict oscillations of the josephson current controlled by the ac phase .
this effect can be used as an alternative sensitive way to detect the ac phase . because the dephasing electrons will not contribute to the current in a josephson junction but will contribute to the current in a normal junction attached to two non - supercoducting leads
, the observed amplitude of the current oscillations due to ac effect in a ring josephson junction should be much larger than the conductance oscillations in the normal ring junction . and
@xmath0 separately on the ring , josephson current can be obtained in this junction . ]
the outline of this paper is as follows . in sec .
ii , we introduce our theoretical model of the josephson current in a 1-d mesoscopic so coupled ring - shaped josephson junction . in sec .
iii , we use the boundary conditions of a multiple terminal junction and present the numerical calculations of the andreev levels and the josephson currents in limits not explored by the analytical solution . in sec .
iv we present an experimental set - up for the observation of the effect and in sec .
v summarize our conclusions . in appendix a , we present the details of the sub - gap states of quasiparticles in superconductor . in appendix
b we show the details of calculating the eigenstates and the eigenenergies of the ring in the tight binding model .
the andreev bound state is a kind of electron - hole bound state in the normal region of the josephson junction depicted by fig .
[ fig1 ] . in the superconductor leads , the excitation spectrum consists of the positive eigenvalues of the bogoliubov equation @xmath1 where @xmath2 is the one - electron hamiltonian , @xmath3 is the electrochemical potential , @xmath4 is the scalar potential , @xmath5 is the superconducting pair potential and @xmath6 is the effective mass . when the excitation energy is less than the gap @xmath7 , the quasiparticles of eq .
( [ bog ] ) in the normal region will be reflected by the pairing potential and form the andreev bound state .
unlike the bound states in a usual square well , the andreev levels carry electric current which contributes to most of the josephson current.@xcite generally speaking , the bound states can be described in such a way that if we have a right moving subgap particle on the left s / n interface fig .
[ fig1 ] , after gaining a phase described by the matrix @xmath8 due to the particle propagation from left to right , reflection on the right interface described by the scattering matrix @xmath9 , a further phase described by the matrix @xmath10 from the motion from right to left , and a reflection on the left interface with the scattering matrix @xmath11 , the state should come back to its original state .
this is true only when @xmath12 , which determines the quantized andreev levels . in zero magnetic field and zero soi , neglecting fermi wavelength mismatch and the barrier on the s / n interface and assuming that the transmission function of electrons in normal region is energy independent , these discrete energies are determined by:@xcite @xmath13,\end{aligned}\ ] ] where @xmath14 is the bcs healing length , @xmath15 is the superconducting phase differences , @xmath16 , @xmath17 is the additional phase due to a point impurity potential @xmath18 in the normal region , @xmath19 is the distance between the point impurity and left interface , and @xmath20 and @xmath21 are corresponding to the transmission probability and reflection probability due to a point impurity when energies are close to the fermi energy .
the term @xmath22 is approximately the phase shift acquired from free electrons and holes propagation in the normal region .
( [ anl ] ) demonstrates that the andreev level will be determined by both the phase shift and the transmission function of electrons and holes in the normal region .
an important limit is when the length of the junction is much shorter than the bcs healing length ( @xmath23 ) so that @xmath24 . in this case , the andreev level takes the form@xcite @xmath25 next we focus on the josephson current in a one - dimentional ( 1-d ) rashba soi ring ; this current can be computed from the superconducting phase dependence of the andreev bound states . for our model calculations
we assume a step - like superconducting pair potential @xmath5 , which is zero in the rashba ring : @xmath26 where @xmath27 are the superconducting phase in the right(left ) leads .
because the soi keeps time - reversal symmetry , the coefficient of the andreev reflection is almost the same as without soi . as a result ,
( [ anl ] ) is still valid qualitatively for a ring - like junction with soi .
the 1d rashba hamiltonian for a 1-channel ring of radius @xmath28 without magnetic field reads:@xcite @xmath29 where @xmath28 is the radius of the ring , @xmath30 with @xmath6 being the effective mass of the electron in the ring , and @xmath31 with @xmath32 being the the strength of the spin - orbit coupling .
the eigenfunctions of this hamiltonian take the form : @xmath33 the associated eigenenergies read @xmath34-\mu , \nonumber \\
e_{e , m}&=&\frac{\hbar \omega_0}{2}[(m_e+\frac{1}{2})^2-(m_e+\frac{1}{2})\sqrt{1+q_r^2}+\frac{1}{4}]-\mu , \nonumber \\ \\ n_{e}&=&\frac{1-\sqrt{1+q_r^2}\pm\sqrt{q_r^2 + 4c_e}}{2},\nonumber \\ m_{e}&=&\frac{\sqrt{1+q_r^2}-1\pm \sqrt{q_r^2 + 4c_e}}{2},\end{aligned}\ ] ] where @xmath35 is the quantized angular momentum number of the electron with spin along @xmath36 direction , @xmath37 is the quantized angular momentum number of the electron with spin along @xmath38 direction , @xmath3 is the chemical potential of the system and @xmath39 is a parameter independent of soi .
since the hole is the time - reversal state of the electron and soi keeps time - reversal symmetry , we have @xmath40 .
therefore , the hole states are the same as the electron states and the associated eigenenergies take the similar form : @xmath41,\nonumber \\ e_{h , m}&=&\mu-\frac{\hbar \omega_0}{2}[(m_h+\frac{1}{2})^2-(m_h+\frac{1}{2})\sqrt{1+q_r^2}+\frac{1}{4}],\nonumber \\ \\
n_{h}&=&\frac{1-\sqrt{1+q_r^2}\pm \sqrt{q_r^2 + 4c_h}}{2},\nonumber\\ m_{h}&=&\frac{\sqrt{1+q_r^2}-1\pm \sqrt{q_r^2 + 4c_h}}{2},\end{aligned}\ ] ] where @xmath42 .
we next consider the phase difference between to the electron and hole propagation in the normal region .
the phase difference in the upper part of the ring for the spin - up and spin - down quasiparticles after being andreev reflected two times take the form @xmath43 and are therefore identical . here
@xmath44 is the phase shift of the quasiparticle with spin along @xmath45 direction .
this identity reflects the fact that there is no spin splitting of the phase shift in this case and is consistent with the time - reversal symmetry of soi .
the phase acquired in the andreev reflection from the interfaces lead to a different phase shift since the spin up electrons and spin down holes have different momenta.@xcite however this phase shift splitting is very small and will not affect the zeros of the josephson current .
therefore we will not consider this splitting in our discussion .
the effect of the rashba interaction on the transmission function has been studied previously , with a conductance oscillation due to the ac phase confirmed experimentally@xcite and theoretically.@xcite under the assumption of a perfect coupling between the leads and the ring and neglecting backscattering , the transmission function takes the form@xcite @xmath46\nonumber \\ &
= & \cos^2[(\sqrt(1+q_r^2)\pi)].\end{aligned}\ ] ] when considering the short junction limit , @xmath47 , the energy @xmath48 of the andreev bound state is affected by the normal - state transmission @xmath20 through eq .
( [ andreev level ] ) .
the josephson current in the low temperature limit takes the form @xcite @xmath49\sin(\phi)}{\sqrt{1-\cos^2[(\sqrt(1+q_r^2)\pi)]\sin^2(\phi/2)}}.\end{aligned}\ ] ] because the transmission probability is affected by the ac phase , an oscillation of the josephson current due to soi should also be observed .
this oscillation will be different from the conductance oscillation because even in the short junction limit , the current is nonlinearly dependent on the transmission function .
although this conclusion is obtained in the short junction limit ( @xmath47 ) , since the zeros of the transmission probability are only dependent on the soi , a similar oscillatin can be expected at any value of @xmath50 .
we will show this to be the case in our numerical calculations .
in this section , we present our numerical investigation on the relation of the josephson current to the soi .
the effetive tight - binding model hamiltonian of the 1d rashba ring is given as @xcite : @xmath51,\nonumber \\ t_{\varphi}^{j , j+1}&=&-i \frac{t_{so}}{(r / a)\delta\varphi}(\cos\varphi_{j , j+1}\hat{\sigma}_x+\sin\varphi_{j , j+1}\hat{\sigma}_y)\nonumber\\ & & + \frac{1}{(r / a)^2(\delta\varphi)^2}t_0 \hat{i}_s,\end{aligned}\ ] ] where @xmath52 , @xmath53 , @xmath54 , @xmath55 , @xmath56 is the radius of the ring and @xmath57 with @xmath19 being the lattice constant of the tight - binding model . the length scale @xmath19 is not related to the lattice constant of the material but is an artifical length scale used in modeling the continum hamiltonian , eq .
( [ 1dham ] ) , and all physical results obtained should be independent of this length scale ; e.g. the fermi energy scales considered should be near the bottom of the tight - binding hamiltonian band , @xmath58 .
the eigenfunctions of this hamiltonian are the same as in eq .
( [ eigen ] ) .
the eigenenergy is obtained as @xmath59\sqrt{1+(\frac{t_{so}}{t_0})^2}\right]-\mu\nonumber \\ & & e_{em}=2t_0\left[1-\cos[(m+\frac{1}{2})\delta\varphi-\beta]\sqrt{1+(\frac{t_{so}}{t_0})^2}\right]-\mu\nonumber\\ & & e_{hn}=-e_{en } , e_{hm}=-e_{em},\end{aligned}\ ] ] where @xmath60 here @xmath61 .
the detailed derivation of the result shown above is presented in appendix b. defining the ac phase as @xmath62 and comparing @xmath63 and @xmath64 in the tight binding model with those in the continuous case , we can find that the term @xmath65 is the counterpart of the term @xmath66 in the continuous case and they can be shown to be equal in the limit @xmath67 . in the following discussion , we will often use the dimensionless parameter @xmath68 instead of @xmath69 in our numerical calculations and plots as a measure of the soi strength since the ac phase is a monotomic increasing function of it .
if we ignore the disoreder in the ring , the phase shift matrices @xmath8 and @xmath10 are diagonal matrices and each diagonal element takes the form @xmath70 .
the s matrices @xmath11 and @xmath9 are calculated by solving the boundary condition at each joint of the superconductor leads and the ring . in a ring system ,
the joint of the lead and the ring can be considered as a 3-way junction .
the boundary condition of this 3-way junction in the tight - binding model is obtained from the principle that the wave function at the joints is also the eigenfunction of the system.@xcite this boundary condition in the tight - binding model allows us to consider more realistic cases such as having different effective mass of electrons in the superconducting leads and normal region .
once s matrices @xmath11 and @xmath9 are obtained , and substituting them into the bound state condition,@xcite @xmath71 we can find the relation among the andreev level @xmath48 , superconductor phase difference @xmath72 and the soi strength .
( [ adbound ] ) contains all of the feyman paths and allows us to consider the effect of multiple backscattering within the ring . in the previous section , to get the analytical result , we made the assumption that the coupling between the ring and the leads are perfect and backscattering was therefore neglected .
however , this idealization is not a good approximation in general since there is usually a very large schottky barrier on the superconductor / semiconductor interface and the backscattering can also strongly affects the electron transmission function through a ring .
all of these factors are considered in our numerical calculation .
first , let us focus on the effects of the backscattering .
we consider the limit @xmath47 . in the tight binding model , @xmath73 .
we assume @xmath74 , @xmath75 ( _ i.e. _ one quarter of the band - width ) , @xmath76 and the perimeter of the ring is @xmath77 which is smaller than the healing coherence length @xmath78 .
we choose @xmath75 because the transmission probability of the electron through the ring oscillates slower in this energy range compared to its zero minimal value and when the energy is close to the bottom of the band . as a result ,
the electron s behavior in this energy range satisfies the approximation we made in the previous section and more directly confirms our analytical discussion .
the parameters corresponding to the experimental system@xcite will be discussed at the end of this section . for different soi strength measured by ac phase(@xmath79 ) .
bottom : andreev levels ( color plot ) vs. superconducting phase difference , @xmath72 , and soi measured by @xmath68 .
here @xmath80 , @xmath75 , @xmath74 , and @xmath76.,title="fig : " ] for different soi strength measured by ac phase(@xmath79 ) .
bottom : andreev levels ( color plot ) vs. superconducting phase difference , @xmath72 , and soi measured by @xmath68 .
here @xmath80 , @xmath75 , @xmath74 , and @xmath76.,title="fig : " ] the andreev levels are presented in fig . [ al1 ] as a function of the superconducting phase difference @xmath72 for different soi strength in the dimensionless variables @xmath81 , @xmath68 , and @xmath82 .
when the ac phase is equal to @xmath83 , the energy of the andreev states is @xmath5 and independent of @xmath72 .
since the josephson current is proportional to the first derivative of the energy @xmath48 of the andreev bound state with respect to the phase difference @xmath72 , this implies that the zeros of the critical current occurs at the values @xmath83 of the ac phase .
this conclusion is consistent with our theoretical result in eq .
( [ andreev level ] ) . in fig .
[ fig4 ] , we show the andreev levels and the normalized josephson currents , @xmath84 , associated with the different ac phases vs. superconducting phase difference .
the results show that the josephson currents amplitudes are tuned by the ac phase . under the short junction limit ,
the energy of the andreev levels is always equal to the superconducting gap @xmath5 when the phase difference @xmath85 and minimum when @xmath86 for any soi strength . therefore the value of @xmath87 is a direct indicator of the amplitude of the critical current .
@xmath88 versus ac phase , @xmath89 , is shown in fig .
the zeros correspond to the value @xmath83 of the ac phase and the peaks are not always corresponding to @xmath90 but show a doubled peaked structure centered around @xmath90 .
this is because of weak localization effect,@xcite similar with the ac phase effect on conductance oscillation in normal junctions.@xcite , and normalized josephson currents ( right ) , @xmath91 , vs. superconducting phase for different soi strength measured by ac phases(@xmath79 ) .
here @xmath80 , @xmath75 , @xmath74 , and @xmath76 . ,
title="fig : " ] , and normalized josephson currents ( right ) , @xmath91 , vs. superconducting phase for different soi strength measured by ac phases(@xmath79 ) .
here @xmath80 , @xmath75 , @xmath74 , and @xmath76 . , title="fig : " ] .
the difference is proportional to the critical current .
here @xmath80 , @xmath75 , @xmath74 , and @xmath76 . ]
we next consider the long channel length limit , @xmath92 , where not only the transmission function but also the phase shift due to the ac effect will affect the andreev levels . as an example , fig .
[ fig7 ] plots the andreev levels as a function of superconducting phase difference @xmath72 with different ac phases in a long - junction . here
the number of sites in the ring is changed from @xmath93 to @xmath94 , _
i.e. _ the length of the ring changes by a factor of 11 , while the other parameters are unchanged .
the andreev levels are shifted due to soi and the josephson current vanishes when @xmath95 as in the case of the short - junction and confirms that the backscattering will not affect the zeros of the critical current in the josephson junction in both the short and long channel length limits .
[ long1 ] for different soi strength measured by ac phase(@xmath79 ) .
here @xmath96 , @xmath75 , @xmath74 , and @xmath76.,title="fig : " ] for different soi strength measured by ac phase(@xmath79 ) . here
@xmath96 , @xmath75 , @xmath74 , and @xmath76.,title="fig : " ] when the schottky barrier and the fermi wavelength mismatch are also considered , the current - soi relation will be more complicated because the electon transmission function will be strongly affected by these factors .
however our numerical resutls show that the zeros of the critical current are the same to the analytical result eq .
( [ jc ] ) .
we explore the josephson current oscillations due to the ac phase at the lower , more realistic , fermi energy @xmath97 , and take @xmath98 , @xmath99 .
the density of the electron in the experiment using hgte hetero - structures varied from @xmath100 to @xmath101 and the effective mass is @xmath102.@xcite to make our parameters match the experimental data , by using the relation @xmath103 , @xmath104 , the parameter @xmath19 is required to be around @xmath105 nm .
when we choose @xmath96 , the perimeter of the ring in our calculation is @xmath106 m which is the same order to the @xmath107 m in the experiment.@xcite the superconducting gap @xmath108 k which is the same order of the gap of the conventional superconductor such as al where @xmath109 k.@xcite to capture the effects of the schottky barrier and the fermi wavelength mismatch on the short and long josephson junction more clear , we explore three ring length scales @xmath110 , @xmath111 , and @xmath112 , shown in fig .
the schottky barrier is chosen to be @xmath113 and the fermi wave vector mismatch is @xmath114 where @xmath115 and @xmath116 are corresponding to the fermi wavevector in the normal region and superconducting leads .
[ short2 ] in different lengtgh scale of the soi ring . here
the fermi energy @xmath97 , the fermi wavelength mismatch between the soi ring and the superconducting leads is @xmath117 and @xmath118.,title="fig : " ] in different lengtgh scale of the soi ring . here
the fermi energy @xmath97 , the fermi wavelength mismatch between the soi ring and the superconducting leads is @xmath117 and @xmath118.,title="fig : " ] in different lengtgh scale of the soi ring . here
the fermi energy @xmath97 , the fermi wavelength mismatch between the soi ring and the superconducting leads is @xmath117 and @xmath118.,title="fig : " ] the above three oscillations have a common characteristic , when the ac phase is equal to @xmath83 , the critical current is zero .
this means the ac effect is present even at a large shottky barrier and fermi wavelength mismatch . in the bottom panel of fig .
[ fig8 ] , the critical current has negative values , which would indicate that in this case there could exists a @xmath119 junction.@xcite this phenomena is due to the large difference of the phase shifts acquired by the electron and hole in the normal region which is related to the nonzero momentum @xmath120 of the cooper pairs @xcite and has been observed in the superconducting / feromagnetic / supconducting junctions.@xcite however the @xmath119 junction can not be realized in our model because it requires the condition @xmath121 and the electron will be totally decohered and the josephson current will not survive .
the advantage of observing the ac effect in the ring josephson junction lays on the fact that the dephasing electrons do not contribute to the josephson current .
the dephasing effect due to the impurities can be considered through the green s function method .
the tunneling hamiltonian used in calculating josephosn current is given by:@xcite @xmath122 where @xmath123 is the annihilation(creation ) operator in left side , @xmath124 is the one in right side , and @xmath125 is the tunneling amplitude of the spin up electron from the left lead with the momentum @xmath126 to the right lead with momentum @xmath127 . in the short junction limit @xmath128
, we can assume @xmath129 to be a constant @xmath20 , and the josephson current is calculated from@xcite @xmath130\end{aligned}\ ] ] where @xmath131 $ ] and @xmath132 is the usual pauli matrix .
this method gives the same analytical form of the josephson current as in eq ( [ andreev level ] ) .
however , when @xmath133 , the parameters of the normal material between the two superconducting leads will affect the tunneling amplitude @xmath129 so strongly that the assumption of the constant @xmath129 is not valid any more .
in this case we write our tunneling hamiltonian as : @xmath134 where @xmath135 is the creation(annihilation ) operator of the electron in the normal material and the index @xmath63 includes all the quantum numbers such as momentum and spin , @xmath136 is the hamiltonian of the normal material , @xmath137 is the eigenvalule of @xmath136 , @xmath138 is the transmission amplitude on the left @xmath139 interface .
@xmath140 in eq .
( [ hamil2 ] ) describes the andreev reflection and can allow us to consider the effect of the phase shifts acquired in the normal region .
the current in this case is modified to read @xmath141,\nonumber \\
t_n = a_{kn}\hat{g}^{r}(r_r , r_l , n),t^*_{n'}=a^*_{kn}\hat{g}^{a}(r_r , r_l , n'),\end{aligned}\ ] ] where @xmath142 is the retarded ( advanced ) green s function of the electron with energy @xmath137 and spin @xmath143 in the normal region . when @xmath144 , @xmath145 is close to the fermi distribution @xmath146 and eq .
( [ gf-+2 ] ) has the same form to its short junction limit eq .
( [ gf-+ ] ) .
however when @xmath147 , because @xmath148 , @xmath149 and @xmath150 are not conjugate to each other . as a result ,
even the product @xmath151 from the trajectories with an identical ensemble of scattering centers , which is corresponding to the ladder correction , is complex with a random phase instead of real .
therefore the ladder correction due to the impurity scattering is zero in the josephson junction current .
this is very different to the calculation of the conductance which is proportional to @xmath152 where the ladder correction due to the impurities is nonzero .
this is why impurity scattering will not affect the josephson current .
this can be confirmed from the experimental data of the fraunhofer - like interference pattern of the critical current in the josephson junctions.@xcite in those experiments , an oscillation with an amplitude of almost @xmath153 of the total current were observed.@xcite similarly in the ring josephson junction with soi , an oscillation due to the ac phase with a larger amplitude than that in the conductance experiments is expected.@xcite especially if a one channel ring limit is achieved , a @xmath153 oscillation due to the ac phase should be observed .
when the normal material is semiconducting , a tunable josephson current controlled by a voltage can be obtained creating a josephson field effect transistor,@xcite where the ring josephson junction can be switched on and off by tuning the ac - phase and carrier concentration .
recently , a voltage of @xmath154 v has been reported to switch off a al / inas / al josephson junction.@xcite our result provides a new possible way to control the josephson current by a gate voltage which can be as small as several mev @xcite and much smaller than the gate voltage in normal josephson field effect transistors.@xcite
in this section , we discuss the feasibility of experiments to observe ac phase through the oscillation of the josephson current . our experimental proposal is based on the radio - frequency ( rf ) method@xcite which is a reliable method to measure the current - phase relation ( c@xmath155r ) of the josephson current@xcite , especially for the small critical current even less than @xmath156 na.@xcite .
therefore this technique makes it possible to detect c@xmath155r in a s / sm / s junction although there is a large schottky barrier on the s / n interface .
our s / sm / s junction , depicted in fig .
[ ring ] , is incorporated into a squid washer of very small and well defined inductance @xmath157 , which is coupled inductively to a high quality tank circuit in resonance with quality @xmath120 and coupled strength @xmath126 fig . [ set ] .
the josephson current @xmath158 in the ring and the phase difference @xmath32 of the dc current and voltage in the tank circuit have the relation @xcite @xmath159 where @xmath160 is the normalized critical current , @xmath161 is the normalized current - phase relation and @xmath162 is the critical current .
therefore , we can observe the oscillation of the josephson current induced by ac phase by detecting the phase difference of the current and voltage in the tank circuit . according to the amplitude of the josephson currents and the size of the josephson junctions in refs .
we can expect a josephson current up to @xmath163 @xmath164a in the ring shape structure whose radius is @xmath165 @xmath164 m and width of each arm is @xmath166 nm . if we choose the inductance of the superconducting loop @xmath167h,@xcite @xmath168 , the normalized critical current @xmath169 , the phase difference @xmath32 of the current and voltage in the tank circuit will be in the range @xmath170 .
in this work we have studied the interplay between the critical josephson current in a ring junction and the ac effect .
we have calculated the andreev levels in a rashba type soi ring system attached to two superconducting leads both analytically and numerically .
numerically , using the boundary condition of the multiple terminal junction in the tight - binding model , we calculate the andreev levels and the josephson currents in both short and long channel length limit . after considering the backscattering in the soi ring , large schottky barrier on the s / n interface
, fermi wavelength mismatch between the superconducting leads and semiconducting ring , and the effect of the impurities in the ring , we find that the oscillations of the josephson current due to the ac phase are robust .
the amplitude oscillation of the josephson current due to the ac effect in the d - c josephson junction is expected to be larger than the conductance oscialltions observed in normal ring structures without superconducting elements .
these results suggest an alternative and likely better way to observe the ac phase .
also , since one period oscillation of the ac phase only needs several mev , we provide a possible way to create josephson junction fets controlled by a voltage of the order of mev .
this work was supported support from onr under grant no .
onr - n000140610122 , by nsf under grant no .
dmr-0547875 , by the research corporation cottrell scholar award , and by swan - nri .
we presents the detail discussion of the quasiparticle states in the gap in this appendix . as we know that the quasiparticle states in the superconductor can be described by the bogoliubov equation @xmath171 where @xmath2 is the one - electron hamiltonian , defined as @xmath172 where @xmath3 is the electrochemical potential
, @xmath4 is the scalar potential and @xmath6 is the effective mass . for a homogeneous superconductor with @xmath173=@xmath174 and @xmath175 ,
the solution of eq.[bogo ] can be written as @xmath176 where @xmath177 , @xmath72 is the phase of the superconductor .
@xmath178 is corresponding to electron - like quasiparticle because in this case , @xmath179 .
for the similar reason , @xmath180 is corresponding to the hole - like quasiparticle . here only the positive energy @xmath48 is considered . in 1-d case , given energy @xmath181 , there are four quasiparticles , two of them are the right moving particles and the other two are the left moving particles . when considering @xmath182 , according to eq.[uv ] , the wave vector @xmath126 satisfies @xmath183 where @xmath184 .
therefor the wave vector @xmath126 has to take the form @xmath185 where the real part and imaginary part of @xmath126 satisfy @xmath186 given energy @xmath182 , the wave vectors of the quasiparticles in this case are corresponding to the four solutions of eq.[wsk2 ] .
the solution of eq.([bogo ] ) will be @xmath187 as a result , @xmath188 . if a normal conductor is coupled to a superconductor , a unique reflection with energy less than gap , namely andreev reflection , can be observed . to calculate the scattering matrix in the interface , we must know the input and output particles , which are corresponding to the right moving and left moving particles separately , in the superconducting lead . it is easy to find them when the energy is larger than the superconducting gap
. however , when the energy is less than the superconducting gap , the wave vector has to be analytic continuous to complex plane eq.[wsk1 ] .
it is hard to say which solution is corresponding the left moving or the right moving particle .
we will figure out this problem by some concepts from retarded green s function .
retarded green s function in the position and energy space , say @xmath189 , is viewed as two traveling wave function outward from a source term @xmath190 .
the opposite case is so called advanced green s function . one way to calculate
s matrix is to use retarded green function@xcite but not advanced green s function because we are interested in the case that if there is an excitation at some point , how the excited particle travels away from the point of the excitation .
this kind of retarded green s function of a fermion is @xmath191 where @xmath192 , therefore we have a pole where @xmath193 . physically we hope that when @xmath194 , the integral gives us the right moving particle and when @xmath195 , the integral gives us the left moving particle .
now let us come back to our question . in our discussion , the energy @xmath48 is chosen to be positive . as a result , @xmath196 must be in the first quadrant of a energy complex plane and just a little bit above the real axis .
actually , eq.([wsk1 ] ) is a general form of wave vector .
@xmath197 is infinitely small when energy @xmath181 and finite when energy @xmath182 .
considering @xmath181 and according to eq.([wsk1 ] ) , the energy of the quasiparticles can be written as @xmath198 here @xmath197 is infinitely small and we neglect the @xmath199 term . since @xmath196 is in the first quadrant of the complex plane , @xmath200 .
for the right moving electron - like quasiparticle @xmath201 , since @xmath202 and @xmath203 , @xmath197 must be also larger than zero , say the wave vector of the right moving electron - like quasiparticle of the first quadrant in complex wave vector plane .
for the similar reason we have the results that the left moving electron - like quasiparticle @xmath204 is in the third quadrant , the right moving hole - like quasiparticle @xmath205 is in the second quadrant and the left moving hole - like quasiparticle @xmath206 is in the fourth quadrant .
although we get this result in the case @xmath207 , it is also valid when @xmath208 because for the wave vectors , the case of @xmath182 is just the analytical continuous of that of @xmath181 . the wave functions of these four quasiparticles within the superconducting gap are @xmath209 where @xmath210 .
an interesting conclusion should be noticed that in both the left and right superconducting leads , exponentially decay quasiparticles are always the input particles and exponentially increase quasiparticles are always output particles .
the velocity of the electrons and the current carried by these electrons can be calculated through the velocity operator and electron current operator @xmath211 where @xmath212 is the charge of an electron .
the velocity of these decay quasiparticles are zero but the current carried by these are not zero .
although eq.([vi ] ) shows that the current is decay , since the quasiparticle will decay to cooper pair@xcite which can carry a supercurrent , the total current due to one quasiparticle is @xmath213
there have been many theoretical papers talking about the rashba interaction in a ring system , such as the eigenenergy , wavefunction and so on by the analytical method in the continuous case and the exact transmission function through the numerical calculation in the tight binding model
. however the eigenenergy and wavefunction in the tight binding model is still undiscussed .
although this is easy to be derived , we write down the conclusion briefly to make our paper more readable .
the eigenfunction of this tight binding model hamiltonian is the same as the eigenfunction of the continuous hamiltonian .
acting the tight binding model hamiltonian eq.[tighth ] on the wave function eq.[eigenf ] and focusing on @xmath214 gives us @xmath217 when @xmath218 , we have @xmath219)\nonumber\\ & & n_{\pm}=\left(\pm\arccos\left((1-\frac{e}{2t_0})/\sqrt{1+(\frac{t_{so}}{t_0})^2}\right)-\beta\right)/\delta\varphi+\frac{1}{2}\\ & & e_{e , m}=2t_0(1-\sqrt{1+(t_{so}/t_0)^2}\cos[(m+1/2)\delta\varphi-\beta])\nonumber\\ & & m_{\pm}=\left(\pm\arccos\left((1-\frac{e}{2t_0})/\sqrt{1+(\frac{t_{so}}{t_0})^2}\right)+\beta\right)/\delta\varphi-\frac{1}{2}\end{aligned}\ ] ] where @xmath61 , @xmath220 and @xmath221 are corresponding to the counterclockwise rotation electron and @xmath222 and @xmath223 are corresponding to the clockwise rotation electron .
we next show the equivalent between the continous limit and the tight bind model by giving the detail proof of our statement that the term @xmath224 is equal to @xmath225 in the limit @xmath226 . according to the l hopital s rule ,
the term @xmath227 satisfies @xmath228 because @xmath229 and @xmath230 .
if we define @xmath231 , we have @xmath232 where @xmath233 @xmath234 by inserting eq.([fa2],[fa3 ] ) to eq.([fa1 ] ) , we find that @xmath235 substituting eq .
( [ fa ] ) to eq .
( [ lhopital ] ) , we obtain the form @xmath236 which is exactly the same as our statement . | we study the interference effect induced by the aharonov - casher phase on the josephson current through a semiconducting ring attached to superconducting leads . using a 1d model that incorporates spin - orbit coupling in the semiconducting ring
, we calculate the andreev levels analytically and numerically , and predict oscillations of the josephson current due to the ac phase .
this result is valid from the point contact limit to the long channel length limit , as defined by the ratio of the junction length and the bcs healing length .
we show in the long channel length limit that the impurity scattering has no effect on the oscillation of the josephson current , in contrast to the case of conductivity oscillations in a spin - orbit coupled ring system attached to normal leads where impurity scattering reduces the amplitude of oscillations .
our results suggest a new scheme to measure the ac phase with , in principle , higher sensitivity .
in addition , this effect allows for control of the josephson current through the gate voltage tuned ac phase . |
recent studies show that , driven by emergence of highly capable devices such as smartphones and resource demanding wireless services such as video streaming , the demand for wireless capacity will increase roughly 1000x , compared to the current 4 g networks , by the year 2020 @xcite . in order to overcome this wireless capacity crunch ,
an evolutionary architecture has been introduced in the next - generation 5 g networks , in which low - cost and lower - power small cell base stations ( sbss ) are densely and randomly deployed within the coverage of macrocell base stations ( mbss ) @xcite and @xcite .
this new architecture has shown its great potential to improve the capacity and coverage of wireless cellular systems .
however , such extreme network densification makes it challenging to manage the cellular system at various levels that include interference control , coverage optimization , load balancing , mobility robustness optimization , and energy management .
many of the existing solutions require autonomous cooperation between network devices .
for example , sbss can cooperate for performing coordinated multipoint ( comp ) transmissions @xcite or for coordinating their interference via techniques such as interference alignment @xcite .
similarly , user cooperation can take place in order to further exploit user diversity , e.g. , user devices can relay for each other by using device - to - device ( d2d ) communications @xcite , or form cooperative groups to support virtual multiple - input multiple - output ( mimo ) @xcite transmissions . despite in emerging wireless networks , most existing optimization techniques
are restricted to addressing the cooperation problem for centralized and homogenous wireless systems in which cooperation is a privilege rather than a necessity . in order to provide a more flexible framework for the cooperative behaviors of future wireless systems , each device in the network can be treated as an individual decision maker that act on its own principles , which naturally leads to game - theoretic approaches where multiple players form a stable and efficient network operating point in a self - organizing manner @xcite .
in particular , the framework of _ cooperative game theory _ provides the necessary tools for modeling and developing self - organizing techniques for forming cooperative groups or coalitions between network devices , based on the mutual benefit and costs for cooperation .
indeed , cooperative games , in general , and coalition formation games ( cf games ) , in particular , have become a popular tool for analyzing wireless networks @xcite . in @xcite , single
antenna users self - organize into multiple coalitions and share their antennas in each coalition to form a virtual mimo system , and hence benefit from spatial diversity or multiplexing . in
@xcite , secondary users ( sus ) form multiple coalitions and combine their individual sensing data at the coalition head to perform cooperative spectrum sensing , and hence improve their sensing performance . in @xcite ,
wireless transmitters form coalitions to coordinate their transmissions so as to improve their physical layer security . in @xcite
, coalition games have been used to overcome the curse of boundary nodes , in which boundary nodes use cooperative transmissions to help the backbone nodes in the middle of the network . in @xcite ,
hedonic coalition formation games are utilized to model the task allocations problem , in which a number of wireless agents are required to collect data from several arbitrarily located tasks using wireless transmissions . in @xcite ,
a coalition formation scheme is proposed for road - side units in vehicular networks to improve the diversity of the information circulating in the network while exploiting the underlying content - sharing vehicle - to - vehicle communication network . in @xcite , a cooperative model based on coalition formation game
is proposed to enable femtocells to improve their performance by sharing spectral resources , minimizing the number of collisions , and maximizing the spatial reuse . in @xcite
, coalition formation games are utilized to strike a balance between the qos provisioning and the energy efficiency in a clustered wireless sensor network .
however , most of this existing body of work focuses on coalition formation models in which the players form separate coalitions and get payoffs from the single coalition they join . in future wireless systems ,
communication nodes are equipped with more powerful devices that are able to utilize multiple resources in a more flexible way .
this makes it possible and necessary to allow nodes to participate in multiple _ overlapping coalitions _ , and , subsequently , receive payoffs from all coalitions they participate @xcite .
for example , a multiple antenna user may join multiple coalitions by devoting its antennas into different groups of users , and benefit from multiple virtual mimo transmissions , and also , an sbs with multiple neighboring sbss needs to coordinate its transmissions with multiple groups of sbss , so as to avoid inner - channel interferences from all its neighbors . in these scenarios ,
the coalitions formed by communication nodes are overlapping , and each node receives payoffs from the multiple coalitions it joins . even though there are many other available modeling tools for analyzing cooperation in wireless networks @xcite , we focus on the methods that belong to cooperative game theory .
the main contribution of this paper is to present an introduction to a novel mathematical framework from cooperative games , _ overlapping coalition formation games _
( ocf games ) , which provides the necessary analytical tools for analyzing how players in a wireless network can cooperate by joining , simultaneously , multiple overlapping coalitions @xcite .
first , in section , we introduce the basic concepts of ocf games in general , and develop two polynomial algorithms for two classes of ocf games .
then , in sections and , based on @xcite and @xcite , we present two emerging applications of ocf games in small cell - based heterogeneous networks ( hetnets ) and cognitive radio networks , in order to show the advantages of forming overlapping coalitions compared with the traditional non - overlapping coalitional games . in section ,
we conclude by summarizing the potential applications of ocf games in future wireless networks .
in this section , we formally introduce the notion of cooperative games with overlapping coalitions , or ocf games . in section .
a , we present the basic model of ocf games and illustrate the overlapping gain compared to traditional cooperative games with non - overlapping coalitions . in section .
b , we focus on one key stability notion in ocf games , @xmath0-core , which is a direct extension of the core from traditional cooperative games . in this regard , we show that the computation of stable " outcomes in the sense of the @xmath0-core can generally be intractable .
therefore , in section .
c , we identify several constraints that lead to tractable subclasses of ocf games , and provide efficient algorithms for solving games that fall under these subclasses .
game theory is a mathematical tool that analyzes systems with multiple decision makers having interdependent objectives and actions .
the decision makers , which are usually referred to as _ players _ , will interact and obtain individual profits from the resulting outcome . in cooperative games
, the players can form cooperative groups , or _ coalitions _
, to jointly increase their profits in a game . in traditional cooperative games ,
the players are typically assumed to form disjoint , non - overlapping coalitions , and they only cooperate with players within the same coalition .
however , there are situations in which some players may be involved in multiple coalitions simultaneously . in such cases
, these players may need to split their resources among the coalitions that they participate in .
for example , a multi - mode wireless terminal may access base stations from different networks and it needs to distribute its traffic load among these networks @xcite . in such situations , some coalitions ( cells of different systems ) may involve some of the same players ( multi - mode terminals ) , and therefore may overlap with one another .
now , we formally introduce the mathematical tool to model these overlapping " situations , _ cooperative games with overlapping coalitions _ , or _
ocf games_. in ocf games , each player possesses a certain amount of resources such as time , power or money . in order to obtain individual profits ,
the players form coalitions by contributing a portion of their resources and receive payoffs from the devoted coalitions .
a _ coalition _ can be represented by the resource vector contributed by its coalition members , i.e. , @xmath1 , where @xmath2 represents player @xmath3 s resources that are contributed to this coalition . for each coalition @xmath4 ,
the _ coalition value _ is decided by a function @xmath5^n \rightarrow \mathbb{r}^+$ ] , which represents the total payoff that the players can get from a cooperative coalition .
the coalition value can be divided to the coalition members based on specific rules , e.g. the value can be equally divided among coalition members , or it can be divided based on the contribution of each member .
we denote by @xmath6 as the payoff allocation rule , and accordingly , the individual payoff that player @xmath3 receives from coalition @xmath4 is denoted by @xmath7 .
the players may decide to devote different amount of resources into different coalitions , so as to maximize its individual payoff @xmath8 , where @xmath9 represents the set of all coalitions @xmath10 formed by the players .
note that these coalitions may have common members , and thus , they form an _ overlapping coalition structure_. compared with the traditional cf games , ocf games allow the players to form an overlapping coalition structure and get payoffs from multiple coalitions .
therefore , the overlapping structure may provide more flexibility for the players to utilize their resources , which enables the coalitions to be better organized and potentially leads to outcomes with higher payoffs .
the following example clearly shows the potential advantage of ocf games .
* example : * consider a software company with three developers @xmath11 and @xmath12 .
each developer works @xmath13 hours a day .
there are two types of projects in the company , big projects and small projects .
a big project requires @xmath14 man - hours per day and provides a @xmath15 bonus , and a small project requires @xmath13 man - hours per day and provides a @xmath16 bonus .
we assume the bonus is divided to the participating developers according to their devoted time . in a traditional cf game
, the players can only form disjoint coalitions , and the optimal coalition structure is @xmath17 , as seen in fig .
[ e1 ] , i.e. , developers @xmath18 and @xmath19 work together to accomplish a big project and developer @xmath12 work alone for a small project .
the total payoffs of @xmath11 and @xmath12 are then given by @xmath20 .
in an ocf game , the players can split their time into different coalitions , and the optimal coalition structure is @xmath21 , as seen in fig .
[ e2 ] , i.e. , developers @xmath18 and @xmath19 devote @xmath13 and @xmath22 hours to accomplish a big project , and developers b and c devote @xmath22 and @xmath13 hours to accomplish another big project .
the payoffs of @xmath11 and @xmath12 are then given by @xmath23 .
in cooperative games , one must seek a stable outcome , i.e. , a coalition structure in which no set of players can deviate and obtain a new structure that increases all their payoffs @xcite . in traditional cf games , the deviating players , or deviators , leave their original coalitions and form a new coalition , the value of which is the total payoff that the deviators can get from their deviation . if there exists a payoff division that makes all the deviators achieve a higher payoff compared with the original coalition structure , we say the deviation is profitable .
if there exists no profitable deviation , we say the structure is stable .
particularly , if the players have an incentive to form the grand coalition that includes all players , the cooperative game is called a canonical game and the set of all stable payoff divisions , corresponding the grand coalition , is called the core of such a game . compared with cf games , defining notions for stability in ocf games is more complicated due to the overlapping property . in ocf games , instead of forming a single coalition , the deviators may form multiple coalitions that overlap one another thus complicating the computation of the maximal total payoff of such an overlapping coalition structure . also , the deviators in ocf games may partially deviate from the original coalitions by withdrawing a portion of their resources while maintaining their other resources in their original coalitions .
therefore , one must precisely define how the non - deviators in the original coalitions will react to such a deviation and how much payoff the deviators can get from those partially deviated coalitions . to this end , we now present the _ @xmath0-core _
, a stability notion suitable for ocf games , which is an extension of the core notion from traditional cf games .
we begin by defining a deviation in ocf games . given a coalition structure @xmath9 and a set of players @xmath24 that attempts to deviate the structure ,
the coalitions in @xmath9 can be divided into two groups : the coalitions that only involve players in @xmath24 , denoted by @xmath25 , and the coalitions that involves players other than @xmath24 , given by @xmath26 . since coalitions
@xmath25 are fully controlled by the deviators , they should be seen as pure resources withdrawn by the deviators . while coalitions @xmath26 , which include both deviators and non - deviators , should be considered as coalitions that are partially deviated by the deviators .
we define @xmath27 as the resources withdrawn from @xmath26 , where @xmath28 is the resources that @xmath24 withdraw from coalition @xmath4 .
the deviators @xmath24 can form an overlapping coalition structure using both the withdrawn resources @xmath29 , and the resources of their own @xmath25 .
we denote by @xmath30 as the sum available resources of the deviators @xmath24 , and @xmath31 as the set of all possible coalition structures that can be formed using @xmath30 . the optimal coalition structure formed by @xmath24
is then given by @xmath32 .
note that the deviators may also receive payoffs from the coalitions that they partially deviate , i.e. , @xmath26 .
we formally define the _ arbitration function _
@xmath33 , which represents the total payoff that the deviators @xmath24 will receive from coalition @xmath34 .
[ a - core ] a deviation on a player set @xmath24 is said to be _
@xmath0-profitable _ if and only if @xmath35 if there exists no @xmath0-profitable deviation for any player set , then the coalition structure @xmath9 is said to be in the _
@xmath0-core _ , or _
@xmath0-stable_. the coalition structures in @xmath0-core represent the stable structures in which no set of players have the motivation to deviate from the current structure .
we note that the definition of @xmath0-core depends on the specific form of the arbitration function .
according to different assumptions on players and different applicable scenarios @xcite , three of the mostly used arbitration functions are described as follows : 1 .
_ c - core _
: if the players are very conservative in cooperating with deviators , the deviators may receive no payoffs from any original coalitions in @xmath26 , even if they still contribute to these coalitions , i.e. , @xmath36 .
this is called the _ conservative arbitration function _ , and the stability notion is referred to as _ c - core_. 2 .
_ r - core _ :
if the players are more lenient , the deviators can still get their original payoffs from the coalitions that are not influenced by their deviation , i.e. , @xmath37 for all coalition @xmath4 with @xmath38 .
this is called the _ refined arbitration function _ , and the stability notion is referred to as _ r - core_. 3 . _ o - core _ : the players can be highly generous that they allow the deviators to keep all the leftover " payoff as long as the non - deviators original payoffs are ensured to be unchanged , i.e. , @xmath39 .
this is called the _ optimistic arbitration function _ , and the stability notion is referred to as _ o - core_. hereinafter , we adopt the _ o - core _ as the stability notion due to its computational advantage that we will explain in the next subsection . to avoid the difficulty in representing the value function @xmath40 and the arbitration function @xmath0 , we assume that the resources are given by integers such that @xmath41 and the players can only divide their resources in a discrete manner .
such games are referred to as _ discrete ocf games _ and they apply to practical systems .
also , due to the cost of information exchange between deviators , we assume the number of deviators in a deviation is preliminarily bounded .
we denote by @xmath42 as the upper bound of deviation size , i.e. , @xmath43 for any deviation on any player set @xmath24 .
it has been shown that computing an @xmath0-stable outcome of an ocf game is generally a challenging problem @xcite . however ,
if we only consider the o - core notion and identify several constraints on the game , there exists efficient algorithms that lead to o - stable outcomes of such games .
[ o - property ] any o - profitable deviation on any player set @xmath44 will increase the _ social welfare _
, which is defined as the total value of all coalitions in the outcome , i.e. , @xmath45
. the social welfare can be written as @xmath46 , where the first item represents the sum value of @xmath47 , and the second item represents the sum value of @xmath48 .
after the deviation of @xmath24 , the deviators withdraw @xmath28 resources from each coalition @xmath49 , and the coalition is reduced to @xmath50 .
meanwhile , the deviators form an optimal coalition structure @xmath51 . therefore
, the social welfare after deviation is given by @xmath52 .
according to the definition of o - core , we can easily prove that the social welfare is increased by the deviation on @xmath24 , i.e. , @xmath53 .
therefore , we conclude that any o - profitable deviation will increase the social welfare . in most practical problems , the social welfare is bounded due to the limited resources . for such games ,
proposition [ o - property ] implies that the game must converge to an o - stable outcome after finite o - profitable deviations .
therefore , we can compute an o - stable outcome by iteratively computing an o - profitable deviation to the current outcome . however , finding an o - profitable deviation is a challenging problem . first
, without further restrictions on deviations , the number of potential deviations can be extremely large .
second , deciding whether a deviation is o - profitable requires solving the optimal coalition structure problem , which is not a straightforward problem .
now , we define two subclasses of ocf games , namely @xmath54-coalition ocf games and @xmath54-task ocf games , and we provide efficient algorithms to compute o - stable outcomes in such games , respectively . in a @xmath54-coalition ocf game , we assume each player can contribute to at most @xmath54 coalitions .
this assumption is reasonable in many practical systems due to geographical constraints , communication cost , or lack of information .
for example , a mobile user can only connect to limited base stations that are close to it .
for a group of deviators @xmath24 in a @xmath54-coalition game , the number of possible deviations is bounded by @xmath55 .
since there are @xmath56 groups of possible deviators , the total number of possible deviations is then bounded by @xmath57 , which is polynomial in @xmath58 . for any deviation on @xmath24 , the deviators need to calculate the optimal coalition structure @xmath51 to decide whether the deviation is o - profitable .
we define the _ superadditive cover _ of @xmath40 to be the function @xmath59^n \rightarrow \mathbb{r}^+$ ] , such that @xmath60 for any resource vector @xmath61 .
briefly , @xmath62 is the maximal total value that the players can generate by forming overlapping coalitions when their total resources are given by @xmath61 .
we observe that @xmath63 , which is a recurrence relation for a discrete - time dynamic system .
thus , we can use the dynamic programming algorithm to calculate @xmath62 .
given the values of @xmath64 for all @xmath65 , the computation of @xmath62 requires @xmath66 times of computing @xmath40 .
therefore , the entire computation of @xmath62 requires at most @xmath67 times of computing @xmath40 .
when @xmath62 is calculated , we can trace backward the optimal path and achieve every coalition in the optimal coalition structure @xmath68 .
therefore , the optimal coalition structure @xmath51 can be calculated in time @xmath69 .
therefore , we can calculate an o - profitable deviation in time @xmath70 , which is polynomial in @xmath58 .
the algorithm for @xmath54-coalition games is shown in table [ k - coalition ] .
p160 mm input an initial outcome @xmath71 . [ 1 ] @xmath72 @xmath73 initial coalition structure @xmath74 @xmath73 initial payoffs @xmath75 for all @xmath34 @xmath76 decide new payoffs @xmath6 @xmath77 @xmath73 final coalitional structure @xmath78 @xmath73 final payoffs output an o - stable outcome @xmath79 .
+ in a @xmath54-task ocf game , each coalition in the game corresponds to a specific task and each player can only contribute to @xmath54 tasks .
being different from @xmath54-coalition ocf games , the number of coalitions in a @xmath54-task ocf game is strictly limited by the number of tasks , which are predetermined by the considered problem .
for example , in a software company , the available projects are predetermined and the developers can not form coalitions to generate new projects but only divide his time among the existing ones .
since the number of coalitions is fixed in a @xmath54-task ocf game , a deviation will not form new coalition structures but only move resources among the existing coalitions , and thus , we refer to deviation as _ transfer _ in @xmath54-task ocf games .
the number of possible transfers is now given by @xmath80^s = \mathcal{o}(n^s)$ ] , which is polynomial in @xmath58 . since the deviators do not form an overlapping coalition structure , their payoffs can be easily calculated using the arbitration function .
therefore , an o - profitable deviation of a @xmath54-task ocf game can be calculated in time @xmath81 .
the algorithm for @xmath54-task games is shown in table [ k - task ] .
p160 mm input an initial outcome @xmath71 .
[ 1 ] @xmath72 @xmath73 initial coalition structure @xmath74 @xmath73 initial payoffs @xmath75 for all @xmath82 decide new payoffs @xmath6 @xmath77 @xmath73 final coalitional structure @xmath78 @xmath73 final payoffs output an o - stable outcome @xmath79 . + given the polynomial algorithms for @xmath54-coalition games and @xmath54-task games , we then provide two example applications to show how the concepts and algorithms of ocf games can be utilized in wireless networks .
note that we restrict our model to single - resource scenarios in which the players only have one type of resources .
however , this model can be extended to the multi - resource setting , by using a vector rather than a scalar to describe the contribution of a player , and all the concepts and algorithms can also be extended to such a case .
in small cell - based hetnets , a large number of small cells may be randomly deployed in the same spectrum as the existing , macro - cellular network . due to the large amount of small cells and their ad hoc nature of deployment ,
interference management is always one of the key challenges in hetnets @xcite .
there are many interference management techniques , such as successive interference cancellation ( sic ) , parallel interference cancellation ( pic ) and multiuser detection ( mud ) @xcite .
however , these techniques require global knowledge of the characteristics of the interfering channels , which generates a huge amount of backhaul traffic for small cells and makes it impractical when the number of small cells is large . in this case , distributed schemes become important , where no central controller is involved in the system and minimum information exchange is required among the small base stations @xcite . game theory , due to its self - organizing characteristic , has also been widely utilized to design distributed approaches for interference management @xcite
. however , most of these approaches are based on non - cooperative games where no information exchange is allowed among small cells . in this paper
, we propose a cooperative approach based on ocf games .
we consider the downlink scenario in which the macro users are interfered by nearby sbss , and small cell users are interfered by mbss as well as nearby sbss . to avoid interference between the macro network and small cells
, the underlayed mbs can inform each sbs in coverage of the available rbs in the current slot , which is determined by the radio assignment of the macro users .
however , interference between small cells still exists due to the lack of coordination between sbss .
this scenario is illustrated in fig .
[ app1 ] . we will study how ocf games can be used to coordinate the interference between sbss and improve the entire network performance @xcite . for each sbs , the available rbs , which are decided by the mbs that covers it ,
are the resources that can be used to cooperate with other sbss . for each available rb ,
the sbs can decide to leave it unoccupied so as to reduce the inter - cell interference , or to utilize it for downlink transmissions so as to improve the throughput .
when the sbss cooperate with each other by contributing a part of their available rbs , they form a coalition in which all contributed rbs are evenly distributed to the involved sbss . to avoid interference inside the coalition , each rb
can only be distributed to one sbs , and the sbss can only utilize the rbs distributed to them .
the value of a coalition is given by the total downlink throughput generated by the resources of this coalition .
note that the interference outside the coalition still exists and it should be considered in the calculation of coalition value . the value is then distributed to the involved sbss as their payoffs when they actually use them in the downlink transmission . in order maximize their individual payoffs
, the sbss may form different overlapping coalitions by deviating from the current overlapping coalition structure .
the dynamics can be seen as an ocf - game , which converges to an o - stable structure as we explained . since the coalitions can only be formed by neighboring sbss ,
the number of coalitions that a sbs can participate is limited .
therefore , the studied ocf game is a @xmath54-coalition ocf game .
we use the developed algorithm in table [ k - coalition ] and the performance is shown in fig .
[ radio ] .
@xmath83 area with different levels of traffic load .
the interference radius of each sbs is set as @xmath84 m.,width=403 ] in fig .
[ radio ] , we compare the developed algorithm with the situation of no overlapping in networks with different levels of traffic load . the values @xmath85 and @xmath86 represent the average rate between the number of required rbs by small cell users and the total available rbs for sbss .
when the sbss are sparsely deployed , the sbss seldom interfere each other , and the performance improvement by the developed algorithm is limited . as more sbss are deployed in the area , the interference coordination becomes crucial and the developed algorithm improves the network performance by @xmath87 to @xmath88 .
when the sbss are extremely dense , the inevitable interference between the coalitions dominates the network performance , and the advantage of the developed algorithm converges to zero .
also , when the traffic load is heavier , interference coordination can bring more benefits to the network , and thus , the developed algorithm performs better .
in order to provide gigabit transmission rate , future wireless networks must use a large amount of spectrum resources .
however , the scarcity of the radio spectrum coupled with the existing , fixed spectrum assignment policies , has motivated the need for new , dynamic spectrum access mechanisms to efficiently exploit the spectral resources .
cognitive radio ( cr ) is one highly promising technique to achieve such dynamic exploitation of the radio spectrum . in cr networks ,
unlicensed , secondary users ( sus ) , can sense the environment and change their parameters to access the spectrum of licensed , primary users ( pus ) , while maintaining the interference to the pus below a tolerable threshold . such spectrum sensing is an integral part of any cr network .
indeed , reaping the benefit of cr is contingent upon deploying smart and efficient spectrum sensing policies . here , we consider the cooperative sensing scheme , in which nearby sus exchange their local sensing results and then make collaborative decisions on the detection of pus .
there are three categories of cooperative spectrum sensing , based on how cooperating sus share their sensing data in the network : centralized , distributed , and relay - assisted @xcite . here , we consider the distributed case , where sus communicate among themselves with no fusion center or relay , as seen in fig . [ app2 ]
traditionally , the distributed approach adopts non - cooperative schemes for its simplicity @xcite . in this paper
, we will show how ocf games can be used to introduce cooperation between sus , and thus , improve the overall sensing performance @xcite .
consider a cognitive radio network with multiple sus equipped with energy detectors and a single pu far away from them . in this network ,
the sus can individually and locally decide on the presence or absence of the pu via their own detectors .
then , they can cooperate with one another by exchanging their local decisions via a reporting channel . at last
, each su combines its local decision with the received decisions and decides whether or not the pu is present .
note that the sus may have different local detectors with different detection threshold , their missed detection probabilities and false alarm probabilities may be different . in process of cooperative
sensing , each su in the system needs to collect local decisions from other sus , and thus , each su represents a sensing task to be accomplished via the cooperation of sus .
however , the bandwidth of the report channel is limited for every su , which is usually not enough to transmit to all other sus , especially for those sus with bad channel conditions .
thus , a su needs to decide how to cooperate with other sus by efficiently allocating its limited bandwidth to the transmissions of different sus .
therefore , this is @xmath54-task ocf game , where each task is presented by a coalition composed of a head su that collects local decisions and other sus that report to it . the value of the coalition is then given by the sensing performance of the head su , which can be calculated based on the adopted fusion rule .
we utilize the developed algorithm in table [ k - task ] and show its performance in fig .
[ cr ] . in fig .
[ cr ] , we compare the developed algorithm with the local sensing method and the non - overlapping method . in the local sensing method ,
the sus use their local sensing decision without any information exchange . in the non - overlapping method , the sus form non - overlapping coalitions and each su only share information inside the coalition it joins .
clearly , the developed algorithm outperforms both methods .
the incorrect probability decreases to @xmath89 of the local sensing method and @xmath90 of the non - overlapping method . also , the improvement becomes larger as the number of sus increases .
ocf game are quite suitable for modeling the future wireless networks , in which the wireless nodes are dense , self - organizing , and cooperative . in this section ,
we briefly discuss other potential applications of ocf games and then summarize the applications in table [ app ] .
* application * & * player * & * resources * & * coalition * & * coalition value * & * type * + radio resource allocation & small cells & available rbs & radio coordination among the rbs from different small cells & total throughput of the coordinated rbs considering all the potential interference & @xmath54-coalition + cooperative spectrum sensing & sus & signaling bits & a specific su and the sus that report to this su & the cooperative sensing performance of this specific secondary user & @xmath54-task + multi - radio traffic offloading & multi - mode devices & user traffic & a specific base station and the traffic contributed by different devices & a function reflecting the user experience & @xmath54-task + comp & base stations & radio resources & a cell - edge user and the resources contributed by different base stations & the throughput of the cell - edge user & @xmath54-task + virtual mimo & mobile phones & radio resources & cooperative users forming a virtual mimo group & the mimo link rate & @xmath54-coalition + smartphone sensing & smartphones & battery energy & a task and the energy devoted by different smartphones & the task utility & @xmath54-task + cellular networks are constantly evolving into their next generation . however , the former systems are not entirely replaced by the new systems .
in fact , it is expected that different networks will coexist for a long time , and , thus , mobile phones will be multi - mode terminals that enable communications over different radio access technologies ( rats ) . in order to fully explore their network investments , the operators must intelligently offload their network traffic over different rats . developing such offloading schemes , which must consider the demands and access authorities of different users , the transmitting rates of different technologies , and the deployment and load of different base stations , is quite challenging for a large number of users and base stations .
however , one can use the proposed @xmath54-task ocf game to model this problem . in the ocf game model
, the mobile users can distribute their traffic into different base stations in different networks .
a coalition here represents a base station as well as the traffic devoted from different mobile users .
the coalition value can be simply defined as the total throughput of this base station with channel and technology limitations , or a sophisticated function reflecting the user experience , which considers the delay and rate experienced by the users , and the cost and energy efficiency of the network . using the developed algorithm in table [ k - task ] ,
the user traffic can be intelligently distributed among different networks with high network performance in the sense of the defined value function . in order to increase the performance of cell - edge users , coordinated multipoint ( comp )
transmission has been proposed , in which the signals of multiple base stations are coordinated to serve a cell - edge user .
since there are multiple cell - edge users , the base stations should allocate their radio resources among these users .
it is a challenging optimization problem , since the channel conditions , traffic demands and radio resources are different for different users and base stations .
however , we can model this problem using a @xmath54-task ocf game . in the ocf game model
, the base stations can freely allocate their radio resources to different users , including bandwidth , power and antenna resources .
a coalition represents a cell - edge user as well as the radio resources devoted from different base stations .
the coalition value is defined as the throughput of this cell - edge user .
thus , using the developed algorithm in table [ k - task ] , the radio resources of base stations can be efficiently distributed among different cell - edge users .
another related application is the cooperation between user devices . in order to increase their transmission rate ,
nearby users may group together to use virtual mimo transmissions .
the mimo link rate is generally increasing with the number of cooperated users , while the marginal increase is decreasing due to the increasing distance between different users .
thus , a user may want to allocate its radio resources among different cooperative groups , so as to maximize its individual throughput .
this problem can be modeled via a @xmath54-coalition ocf game , in which a coalition represents a virtual mimo group and the coalition value is the mimo link rate . using the developed algorithm in table [ k - coalition ] , the radio resources of users can be efficiently distributed among different virtual mimo groups . in recent years , smartphones are equipped with more and more sensors .
these powerful sensors allow public departments or commercial companies to accomplish large - area sensing tasks via individual smartphones .
these tasks often require collecting data in a large area , and thus , a huge number of smartphones may be involved .
based on the task itself and the geographic locations of smartphones , different tasks may require different amount of energy and provide different payoffs for different smartphones . a smartphone user must decide to which tasks he should devote the limited energy .
therefore , we can model this problem with the studied @xmath54-task ocf game , in which each coalition represents a task and the energy devoted from different smartphones , and the coalition value is given by the task utility . using the developed algorithm in table [ k - task ]
, the smartphone users can efficiently allocate their energy into different sensing tasks .
in this paper , we have introduced the framework of overlapping coalition formation games as a tool to model and analyze the communication scenarios in future networks .
in particular , we have defined two subclasses , namely @xmath54-coalition and @xmath54-task ocf games , and we have developed polynomial algorithms to achieve an o - stable outcome .
subsequently , we have presented , in detail , how ocf games can be used to address challenging problems in two application domains : radio resource allocation in hetnets and cooperative sensing .
in addition , we have discussed some other potential applications of ocf - games , including multi - radio traffic offloading , cooperative communications , and smartphone sensing . finally , we envision that the use of the ocf game framework will play an important role in 5 g networks , particularly , as the network becomes more dense , decentralized and self - organizing .
o. onireti , f. heliot , and m. a. imran , on the energy efficiency - spectral efficiency trade - off in the uplink of comp system , " _ ieee transactions on wireless communications _ , vol .
11 , no . 2 , pp . 556 - 561 ,
f. pantisano , m. bennis , w. saad , m. debbah , and m. latva - aho , interference alignment for cooperative femtocell networks : a game - theoretic approach , " ieee transactions on mobile computing , vol .
12 , no . 11 , pp . 2233 - 2246 , nov .
y. kawamoto , j. liu , h. nishiyama , and n. kato , an efficient traffic detouring method by using device - to - device communication technologies in heterogeneous network . " , _ ieee wireless communications and networking conference _ , istanbul , turkey , apr .
2014 . w. feng , y. wang , n. ge , j. lu , and j. zhang , virtual mimo in multi - cell distributed antenna systems : coordinated transmissions with large - scale csit , " _ ieee journal on selected areas in communications _
31 , no .
10 , pp . 2067 - 2081 , oct .
2013 .
w. saad , z. han , m. debbah , a. hjorungnes , and t. basar , coalitional game theory for communication networks , " _ ieee signal processing magazine , special issue on game theory _ ,
26 , no . 5 , pp .
77 - 97 , sep . 2009 .
w. saad , z. han , m. debbah , and a. hjorungnes , a distributed merge and split algorithm for fair cooperation in wireless networks , " in _ proceedings of international conferernce on communications , workshop cooperative communications and networking _ , beijing , china , may 2008 .
w. saad , z. han , m. debbah , a. hjorungnes , and t. basar , coalitional games for distributed collaborative spectrum sensing in cognitive radio networks , " in _ proceedings of ieee infocom _ , rio de janeiro , brazil , apr .
w. saad , z. han , t. basar , m. debbah , and a. hjorungnes , physical layer security : coalitional games for distributed cooperation , " in _ proceedings of 7th international symposium of modeling and optimization in mobile , ad hoc , and wireless networks _ ,
seoul , south korea , jun .
z. han , h. v. poor , coalition games with cooperative transmission : a cure for the curse of boundary nodes in selfish packet - forwarding wireless networks , " _ ieee transactions on communications _
57 , no . 1 , pp . 203 - 213 , jan . 2009 .
w. saad , z. han , t. basar , m. debbah , a. hjorungnes , hedonic coalition formation for distributed task allocation among wireless agents , " _ ieee transactions on mobile computing _
10 , no . 9 , pp . 1327 - 1344 , dec . 2010 .
w. saad , z. han , a. hjorungnes , d. niyato , and e. hossain , coalition formation games for distributed cooperation among roadside units in vehicular networks , " _ ieee journal on selected areas in communications _
29 , no . 1 ,
48 - 60 , jan . 2011 .
f. pantisano , m. bennis , w. saad w , r. verdone , m. latva - aho , coalition formation games for femtocell interference management : a recursive core approach , " in _ proceedings of wireless communications and networking conference _
, quintana - roo , mexico , mar .
d. wu , y. cai , l. zhou , and j. wang , a cooperative communication scheme based on coalition formation game in clustered wireless sensor networks , " _ ieee transactions on wireless communications _
11 , no . 3 , pp .
1190 - 1200 , feb . 2012 .
z. zhang , l. song , z. han , and w. saad ,
coalitional games with overlapping coalitions for interference management in small cell networks , " _ ieee transactions on wireless communications _ , vol . 13 , no . 5 , pp .
2659 - 2669 , may 2014 .
t. wang , l. song , z. han , and w. saad , overlapping coalitional games for collaborative sensing in cognitive radio networks , " in _ proceedings of wireless communications and networking conference _ ,
shanghai , china , apr .
2013 . b. di , t. wang , l. song , and z. han , incentive mechanism for collaborative smartphone sensing using overlapping coalition formation games , " in _ proceedings of global communications conference _ ,
atlanta , ga , dec . 2013 .
m. bennis , m. simsek , w. saad , s. valentin , m. debbah , and a. czylwik , when cellular meets wifi in wireless small cell networks , " _ ieee communications magazine , special issue on heterogeneous networks _ ,
vol . 51 , no . 6 , jun . 2013
. p. sangkyu and b. saewoong , dynamic inter - cell interference avoidance in self - organizing femtocell networks , " in _ proceedings of ieee international conference on communications _ ,
tokyo , japan , jun .
2011 .
y. zick , g. chalkiadakis , and e. elkind , overlapping coalition formation games charting the tractability frontier , " in _ proceedings of 10th international conference on autonomous agents and multiagent systems _ , valencia ,
spain , jun .
x. chen , x. gong , l. yang , and j. zhang , a social group utility maximization framework with applications in database assisted spectrum access , " in _ proceedings of ieee infocom _ ,
toronto , canada , apr . | modern cellular networks are witnessing an unprecedented evolution from classical , centralized and homogenous architectures into a mix of various technologies , in which the network devices are densely and randomly deployed in a decentralized and heterogenous architecture .
this shift in network architecture requires network devices to become more autonomous and , potentially , cooperate with one another
. such cooperation can , for example , take place between interfering small access points that seek to coordinate their radio resource allocation , nearby single - antenna users that can cooperatively perform virtual mimo communications , or even unlicensed users that wish to cooperatively sense the spectrum of the licensed users .
such cooperative mechanisms involve the simultaneous sharing and distribution of resources among a number of overlapping cooperative groups or coalitions . in this paper , a novel mathematical framework from cooperative games , dubbed _ overlapping coalition formation games _ ( ocf games ) , is introduced to model and solve such cooperative scenarios .
first , the concepts of ocf games are presented , and then , several algorithmic aspects are studied for two main classes of ocf games .
subsequently , two example applications , namely , interference management and cooperative spectrum sensing , are discussed in detail to show how the proposed models and algorithms can be used in the future scenarios of wireless systems .
finally , we conclude by providing an overview on future directions and applications of ocf games . |
in virtue of parity violation the experiments have shown that all fermions emitted from decay processes are longitudinally polarized .
it is well known that neutrinos are left - handed ( lh ) polarized while antineutrinos are right - handed ( rh ) polarized
. in beta or muon decays it is found that electrons are lh polarized while positrons are rh polarized.@xmath0}$ ] based on these experimental facts , naturally , it is thought that not only the final - state fermions , but also the initial - state fermion should reveal the feature of longitudinal polarization in weak interactions .
so we could further consider that the lifetime of fermions in the lh helicity state and that in the rh helicity state should be different .
namely , the lifetime asymmetry should exist in left - right handed polarized fermions .
however , the experimental and theoretical study on the polarization or helicity of the initial fermions in decays has not yet been discussed fully in the literature . in a previous paper@xmath1}$ ] we have given a first discussion of the problem on lifetime asymmetry .
in this paper we continue the study of this issue .
the differences among spin states , helicity states and chirality states are emphasized and the lifetime asymmetry of polarized muons will be calculated concretely .
in the standard model ( sm ) , all of fundamental fermions are divided into two classes , lh chirality state and rh chirality state , in order to describe the parity violation in weak interactions .
the lh chirality state is different from the rh chirality state .
the former is a su(2)-doublet state whereas the latter a su(2)-singlet state , hence they have different gauge transformations .
especially , rh chirality states have zero weak isospin and are only present in neutral weak currents .
therefore , only lh chirality states exist in charged weak currents .
the interaction lagrangians for charged weak lepton current and charged weak quark current read , respectively @xmath2 where @xmath3 is the coupling constant corresponding to su(2 ) and the subscript @xmath4 denotes the lh chirality state . obviously , all fermions are in the lh chirality states while all antifermions are in the rh chirality states .
for example , the weak interaction in muon decay is successfully described by four - fermion interaction hamiltonian .
we denote the matrix element by @xmath5 where @xmath6 indicates a scalar , vector or tensor interaction , and @xmath7 indicate a right- or left - handed chirality of the electron or muon .
the chiralities @xmath8 and @xmath9 of the @xmath10 and @xmath11 are then determined by the values of @xmath12 and @xmath13 .
all coupling constants have been obtained entirely from experiments without any model assumption.@xmath14}$ ] the experiments on muon decay show @xmath15 , @xmath16 , @xmath17 to be zero , and at least one of the two coupling , @xmath18 or @xmath19 , to be nonzero .
the experiments on inverse muon decay provide a lower limit for pure @xmath20 interaction with @xmath21 .
thus the measurements give a strong support to the sm which sets @xmath22 while all others being zero , and then indicate that the charged weak current is dominated by a coupling to left - handed chirality fermions .
therefore , the negative muon decay can be written as @xmath23 and the matrix element ( 3 ) has the form @xmath24
there exist three kinds of spinor wave functions , i.e. , the spin states , the helicity states and the chirality states .
we will discuss the difference and the relation among them .
the spin states are the plane wave solutions of dirac equation , and in momentum representation for a given four - momentum @xmath25 and mass @xmath9 , the positive energy solution and the so - called negative energy solution are respectively @xmath26 @xmath27 where @xmath28 and @xmath29 are pauli spin wave functions .
the state with @xmath30 is spin up ( @xmath31 ) while the state with @xmath32 is spin down ( @xmath33 ) .
they are eigenstates of operator , @xmath34 , with eigenvalues @xmath35 , namely @xmath36 here @xmath37 is the pauli - lubanski covariant spin vector and @xmath38 is the four - polarization vector in the form @xmath39 which is normalized @xmath40 , orthogonal to @xmath41 . in the rest frame
@xmath38 reduces to @xmath42 . in application
it is frequently necessary to evaluate spin sums in the form @xmath43 the @xmath44 and @xmath45 are the positive energy projection operator and the negative energy projection operator , @xmath46 respectively and @xmath47 are spin projection operators : @xmath48 the plus sign refers to @xmath30 and the minus sign to @xmath32 .
one sees that the operator @xmath49 project out the positive energy states with spin up and @xmath50 the positive energy states with spin down , whereas the operator @xmath51 the negative energy states with spin down and @xmath52 the negative energy states with spin up in its rest frame .
a helicity state is the eigenstate of the helicity of fermions and satisfies the ordinary dirac equation@xmath53}$ ] .
if spinor @xmath54 is taken as the eigenstate of the spin component along the direction of its motion , @xmath55 then the helicity states read @xmath56 and @xmath57 where @xmath58 and @xmath59 are the polar angles of momentum @xmath60 in polar - coordinates .
the state with @xmath61 is the rh helicity state while the state with @xmath62 is the lh helicity state .
because a helicity state is also a plane wave solutions of dirac equation , the energy projection operator of helicity state is equal to that of spin state , namely @xmath63 the projection operators of the helicity states are@xmath64}$ ] @xmath65 we can see that the helicity state is entirely different from the spin state .
both the helicity eigenvalue and the projection operator of helicity state are not lorentz invariant , and the latter is essentially a two - component operator . from eqs .
( [ eq:6 ] ) , ( [ eq:7 ] ) and ( [ eq:8 ] ) we can find out that a spin state with the same @xmath66 but different values of @xmath67 is helicity degenerate .
so the spin states can not uniquely describe the helicity of fermions . on the other hand , taking the simplest case of @xmath68 , which does not lose the universality of problem , we have the lh helicity state @xmath69 and the rh helicity state @xmath70 , respectively @xmath71 even so comparing eq .
( [ eq:12 ] ) with eq .
( [ eq:15 ] ) one can also see that the projection operator of spin state is different from that of helicity state though the spin and helicity state are formally identical when @xmath72 .
furthermore , we should point out emphatically that the helicity and the degree of polarization are two different concepts .
the helicity @xmath67 describes the mutual relation between the spin direction of fermions and its momentum direction .
because helicity changes its sign under the space inversion , it is called a pseudoscalar .
the degree of polarization is the length of polarization vector @xmath73 which is the ensemble average of spin vector of fermion beam .
the polarization vector does not change its sign under the space inversion and so is called a pseudovector or an axial vector , like angular momentum .
the helicity can only take the value of either @xmath74 or @xmath75 and has no meaning in its rest frame .
but the degree of polarization can take any value between @xmath76 and @xmath74 and the value remains constant in different frame . in the spin states the polarization of fermions is described by polarization vector .
it is the discrepancy between the helicity and the polarization vector that results in the spin state being different from the helicity state .
the four - polarization vector @xmath38 is the relativistic generalization of three - polarization vector @xmath73 .
it is able to prove from eq .
( [ eq:12 ] ) that the vector @xmath77 is the ensemble average of spin vector @xmath78 , i.e. , @xmath79 on the other hand , however , we can see out from eq .
( [ eq:9 ] ) that the property of vector @xmath77 is different from that of pseudovector @xmath73 .
for example , when @xmath80 vector @xmath77 reads @xmath81 differing from vector @xmath73 , obviously , the direction of vector @xmath77 is always pointing to @xmath82 axis and its value can be greater than one , i.e. , @xmath83 .
strictly speaking , only in the rest frame can the four - polarization vector @xmath38 be most unambiguously defined@xmath84}$ ] . and the spin projection operators @xmath47 , which are lorentz invariant , can only project out the states which in its rest frame have spin @xmath30 and @xmath85 , respectively .
therefore , the pauli - lubanski covariant spin vector @xmath37 and the four - polarization vector @xmath38 , not like four - momenta , have no intuitively and definitely physical significance in the motion frame .
we reach a conclusion that the polarization of fermions must be described by the helicity states which are closely related to directly observable quantity experimentally .
the chirality states are the eigenstates of chirality operator @xmath86 .
the lh chirality state and the rh chirality state are defined as , respectively @xmath87 in general , chirality states are different from helicity states .
only if @xmath88 ( for example neutrinos ) or @xmath89 ( in the ultrarelativistic limit ) the fermions satisfy weyl equation@xmath90}$ ] , the spinor @xmath29 must then be taken to be eigenstates of helicity operator @xmath67 and the polarization is always in the direction of motion@xmath91}$ ] . in other words , for @xmath88 the helicity states , the chirality states and spin states are identical , i.e. @xmath92 the superscript @xmath93 refers to it being a solution of weyl equation
. the helicity states @xmath69 and @xmath70 in eqs .
( [ eq:16 ] ) and ( [ eq:17 ] ) can be expanded as linear combination of chirality states , respectively@xmath94}$ ] @xmath95 where @xmath96 and @xmath97 are chirality states in the rest frame , @xmath98 the coefficients @xmath99 , @xmath100 , @xmath101 and @xmath102 as given by @xmath103 where @xmath104 is the velocity of the muons .
it is obvious from eqs .
( [ eq:21 ] ) and ( [ eq:22 ] ) that in a lh helicity state @xmath105 the coefficient @xmath99 is the amplitude of lh chirality state @xmath106 and the @xmath100 that of rh chirality state @xmath107 ; while in a rh helicity state @xmath108 the @xmath101 that of lh chirality state @xmath109 and the @xmath102 that of rh chirality state @xmath110 in its rest frame . for lh helicity state the hidden amplitude of rh chirality state decreases with the increase of @xmath104 until @xmath111
when @xmath112 , showing that a high - energy fermion can be lh polarized without hidden rh spinning instability . for rh helicity state the hidden amplitude of lh chirality state decreases with the increase of @xmath104 .
when @xmath113 , the hidden amplitude of lh chirality state is equal to that of rh chirality state and one even can not discriminate a rest fermion being either lh or rh polarized@xmath114}$ ] .
now let us consider a @xmath115 decay process ( 4 ) .
the lowest order decay rate or lifetime @xmath116 for muon decays , based on the perturbation theory of weak interactions , is given by @xmath117 if the muons are unpolarized and if we do not observe the polarization of final - state fermions , then the transition matrix element , eq .
( [ eq:5 ] ) , is given by averaging over the muon spin and summing over all final fermion spins : @xmath118 ^ 2\nonumber\\ & & \times\left[\;\overline{u}_r(k)\;\gamma_\lambda\;(1+\gamma_5)\;u_s(p)\;\right]^2.\end{aligned}\ ] ] where @xmath25 , @xmath119 , @xmath120 and @xmath121 are 4-momenta , while @xmath66 , @xmath122 , @xmath123 and @xmath124 are spin indices for @xmath13 , @xmath38 , @xmath125 and @xmath126 , respectively .
for the convenience of discussion below , in eq .
( [ eq:27 ] ) we set @xmath127 ^ 2,\ ] ] which is related to the muons . by means of eqs .
( [ eq:10 ] ) and ( [ eq:12 ] ) , the evaluations of spin sums are reduced to the calculation of projection operators : @xmath128 one sees that the explicit evaluation of spin projection operators disappear .
applying eqs .
( [ eq:29 ] ) , ( [ eq:30 ] ) and ( [ eq:11 ] ) as well as the trace theorems we obtain @xmath129 substituting eq .
( [ eq:31 ] ) into eq .
( [ eq:26 ] ) , one has @xmath130 where @xmath131 obviously decay amplitude @xmath132 is a lorentz - invariant matrix element .
therefore we have @xmath133 where @xmath134 is @xmath132 in the muon rest frame .
it is easy to see from eq .
( [ eq:32 ] ) that the integration to the right of @xmath135 is lorentz invariant .
neglecting electron mass , the muon lifetime @xmath136 in its rest frame is given by @xmath137 where @xmath138 is muon mass . in an arbitrary frame the muon lifetime is given by @xmath139 we see that the lifetime is proportional to the velocity @xmath104 of muons as required by special relativity@xmath140}$ ] . so that the lifetime is not a lorentz scalar . for polarized muons
the muon spin should not be averaged . in most literatures and textbooks@xmath141}$ ]
the polarized states of fermions were usually expressed by the spin states ( [ eq:6 ] ) and ( [ eq:7 ] ) . instead of eqs .
( [ eq:29 ] ) and ( [ eq:30 ] ) , we have @xmath142 ^ 2,\ ] ] and @xmath143 respectively . substituting eq .
( [ eq:36 ] ) into ( [ eq:35 ] ) and eq .
( [ eq:35 ] ) into ( [ eq:28 ] ) we obtain @xmath144 where @xmath145 where @xmath146 is the decay amplitude related to the polarization vector @xmath38 : @xmath147 obviously , decay amplitude @xmath148 is also a lorentz scalar , like @xmath132 .
therefore we have @xmath149 where @xmath150 is @xmath148 in the muon rest frame . in a similar way to eq .
( [ eq:32 ] ) , we obtain @xmath151 one easily verifies that the second integration over @xmath146 vanishes , i.e. , @xmath152 .
the first integration is identical with eq .
( [ eq:32 ] ) and we obtain @xmath153 it is also easy to see in the above calculation that the lifetime in the laboratory frame does not exhibit any lifetime asymmetry when the polarization of muons is described by spin states . as mentioned above , however , this method does not enable us to discuss the dependence of lifetime on the polarization of muons . because a spin state is helicity degenerate , the polarization of muons must be described by helicity states . for lh polarized muons , substituting the spin states in eq .
( [ eq:35 ] ) with the helicity states and considering eq .
( [ eq:15 - 1 ] ) , we obtain @xmath154 ^ 2\nonumber\\ & = & \sum_{r=1}^2\left[\overline{u}_r(k)\gamma_\lambda ( 1+\gamma_5)u_{_{lh}}(p)\right]^2.\label{eq:44}\end{aligned}\ ] ] from eqs .
( [ eq:21 ] ) , ( [ eq:24 ] ) and ( [ eq:18 ] ) we easily find @xmath155 where @xmath156 is the spin state in the muon rest frame .
one can see that the chirality - state projection operator @xmath157 picks out lh chirality state @xmath106 in a lh helicity state , which is factorized into two parts in the second equation , one is the spin state @xmath156 and another is a factor @xmath158 which depends on muon s helicity . substituting eq .
( [ eq:45 ] ) into eq .
( [ eq:44 ] ) we have @xmath159 ^ 2\ ] ] comparing eq .
( [ eq:46 ] ) with eq .
( [ eq:35 ] ) and considering eq .
( [ eq:40 ] ) we find out the decay amplitude of lh polarized muons @xmath160 then the lh polarized fermion lifetime is given by @xmath161 similarly , for rh polarized muons we obtain @xmath162 ^ 2\nonumber\\ & = & ( 1-\beta)\sum_{r=1}^2\left[\overline{u}_r(k)\gamma_\lambda ( 1+\gamma_5)u^0_1\right]^2\end{aligned}\ ] ] and the decay amplitude is @xmath163 then the rh polarized fermion lifetime is given by @xmath164 comparing eqs .
( [ eq:47 ] ) and ( [ eq:50 ] ) with eq .
( [ eq:38 ] ) , respectively and considering eqs .
( [ eq:41 ] ) , ( [ eq:41 - 1 ] ) and ( [ eq:43 ] ) , we find the polarized muon lifetime @xmath165 it is easy to see that the @xmath166 is greater than @xmath167 , which shows the lifetime asymmetry of left - right handed polarized fermions .
the lifetime asymmetry is expressed by @xmath168 when @xmath113 , we find @xmath169 ; when @xmath170 , we find @xmath171 , @xmath172 . in particular , when @xmath173 , the lifetime of the lh polarized fermions has a minimum value @xmath174 , as shown in fig .
[ fig : lifetime ] .
lifetime as a function of muon velocity @xmath104 .
( a ) the lifetime @xmath166 of right - handed polarized muons .
( b ) the lifetime @xmath167 of left - handed polarized muons .
( c ) the lifetime @xmath116 of unpolarized muons . ]
it is not difficult to prove that the lifetimes of antimuons @xmath175 in flight are given by @xmath176
the calculation has established that the lifetime of rh polarized muons is different from that of lh polarized muons in flight . under a space reflection the lh helicity state eq .
( [ eq:21 ] ) and the rh helicity state eq .
( [ eq:22 ] ) transform to each other , therefore the set of equations eq .
( [ eq:52 ] ) and eq .
( [ eq:54 ] ) are valid all the same , respectively .
furthermore , this conclusion is also valid for all fermions in the decays under weak interactions .
it means that the @xmath167 is always smaller than @xmath166 for fermions and the @xmath177 smaller than @xmath178 for antifermions in any one of inertial systems in which fermions or antifermions are in flight with a same speed .
the lifetime asymmetry shows a maximum violation of parity symmetry .
hence under the condition of parity violation the lifetime is neither a four - dimensional scalar , nor a scalar under the three - dimensional space inversion .
we emphasize here an important concept that a spin state is helicity degenerate and the spin projection operators @xmath47 can only project out the spin states , but can not project out the helicity states .
the above calculation shows that the so - called eigenstate of operator given by eq .
( [ eq:8 ] ) is by no means a helicity eigenstate ( even we had chosen @xmath60 vector along z axis ) and this is why the parity violation result , eqs .
( [ eq:52 ] ) and ( [ eq:54 ] ) , was overlooked in the past for so long a time even one did not perform the spin average for muons in the laboratory frame .
therefore , the polarized fermions must be expressed by the helicity states which are relevant to physical interpretation and experimental test .
if we find out nonzero pseudoscalar of similar @xmath179 in a physical process , then parity conservation law is certainly violated and lorentz invariance is also violated because @xmath179 is not a lorentz scalar . in the special theory of relativity space coordinates
are treated on an equal footing with time coordinate .
the violation of pure space inversion , in some sense , implies a violation to the equal status and mutual transformation between space and time . and
it in turn implies that something might go beyond the theory of special relativity@xmath180}$ ] .
the lifetime asymmetry is just such an example .
it poses a serious challenge to beautiful symmetry in physics .
when the polarization of muons is described by spin states , it can be seen from eq .
( [ eq:39 ] ) that the factor related to polarization reads @xmath181 the pseudoscalars @xmath182 and @xmath183 are included in it , which reveals the asymmetric distribution of electrons and neutrinos emitted from decaying muons and exhibits parity nonconservation in the final channels .
the property @xmath184 ensures the pseudoscalar @xmath185 , helicity operator , does not appear in the decay amplitude .
therefore , the decay amplitude corresponding to the spin states is independent of muon helicity , which is consistent with the helicity degeneracy of the spin states . differing from the spin states , however , for massive muons lh and rh helicity state may transform to each other as observer s velocity paralleling and exceeding the muon s one . if , therefore , the decay amplitude corresponding to the helicity state were a lorentz scalar , then the decay amplitude of lh polarized muons would be equal to that of rh one and starting from eqs .
( [ eq:47 ] ) , ( [ eq:50 ] ) and ( [ eq:38 ] ) , we would obtain @xmath186 obviously , there exists no any pseudoscalars in it , which shows that the parity is conserved and is clearly wrong .
hence we conclude that the decay amplitude @xmath187 and @xmath188 corresponding to the helicity states must be not lorentz invariant , as shown eqs .
( [ eq:47 ] ) and ( [ eq:50 ] ) .
it is the root cause of the lifetime asymmetry . in other words ,
if the polarization is described by the helicity states , then the lifetime asymmetry will be inevitable .
of course , the average value of decay amplitude , @xmath189 , is still a lorentz scalar . in the before ,
physicists too believed in the covariant form of four spinor which enjoys the invariance under the lt and so misunderstood the meaning of so - called spin states .
meanwhile , the importance of helicity state and its difference from spin state and chiral state were often confused or overlooked .
our lesson and experience are just focused on the above crucial point before we are able to get rid of the constraint of covariant four spinor form .
what we eventually realized is to discriminate two things : while a physical law like that reflected by the dirac equation is one thing , a phenomenon like a particle s velocity , its helicity as well as its decay lifetime is another thing .
a law should be invariant under the lt , but a phenomenon needs not .
actually , a phenomenon may be different in different inertial systems .
in short , the information is not existing totally in the sense of objectiveness , it is created by the subject and object in common . the measurements on muon decay used to be performed in its rest frame .
it was realized that muons , formed by forward decay in flight of pions inside cyclotron , were stopped in a nuclear emulsion , sulphur , carbon , calcium or polyethylene target .
the polarization effects of muon decay were observed using carbon stopping target@xmath190}$ ] , in which there is no depolarization of the muons .
so far the measurement of the lifetime of polarized muons in flight has not yet been found in literature .
therefore , one actually lacks any direct experimental evidence either to support or to refute the lifetime asymmetry .
we report it here now in the hope that it may stimulate and encourage further experimental investigations on the question of the lifetime asymmetry in muon decays .
15 h. frauenfelder , r. bobone and e. v. goeler , _ et al .
_ , phys .
rev . * 106 * , 386 ( 1957 ) ; d. h. perkins , _ introduction to high energy physics _ ( reading , massachusetts : addison - wesley publishing company , 1972 ) , pp.135 - 150 .
particle data group , the europ .
j. * c15 * , 316 - 318 ( 2000 ) .
z. q. shi and g. j. ni , _
arxiv : hep - ph/0202017 . + z. q. shi and g.
j. ni , chin .
* 19 * , 1427 ( 2002 ) .
y. kuno and y. okada , rev .
73 * , 151 ( 2001 ) ; w. fetscher , h - j . gerber and k. f. johnson , phys .
* b173 * , 102 ( 1989 ) ; charm ii collaboration , p. vilain , _ et al .
_ , phys .
b364 * , 121 ( 1995 ) . v. v. dvoeglazov , _ preprint _ arxiv : math - ph/0309002 . c. itzykson and j. b. zuber , _ quantum field theory _
( new york : mc graw - hill book company , 1980 ) , p. 59 .
h. frampton and w. k. tung , phys .
* d3 * , 1114 ( 1971 ) . g. j. ni and t. chang , journal of shaanxi normal university ( natural science edition ) * 30*(3 ) , 32 ( 2002 ) ; _ preprint _ arxiv : hep - ph/0103051 .
d. lurie , _ particles and fields _
( new york : john wiley & sons , 1968 ) , p. 25 . z. q. shi and g. j. ni , _ preprint _ arxiv : hep - ph/0211051 ; accepted by relativity , gravitation , cosmology , 2004 .
l. h. ryder , _ quantum field theory _
( cambridge : cambridge university press , 1996 ) . j. d. bjorken and s. d. drell , _ relativistic quantum mechanics _ ( new york : mc graw - hill book company , 1964 ) , p. 263 .
d. bailin , _ weak interaction _
( adam hilger , bristol , 1982 ) , chapter 3 ; y. kuno and y. okada , rev .
phys . * 73 * , 158 ( 2001 ) .
g. j. ni , _ preprint _
arxiv : hep - ph/0306028 .
g. bardin , j. duclos and a. magnon , _ et al .
_ , phys .
* b137 * , 135 ( 1984 ) . | based on the parity violation in standard model , we study the dependence of the lifetime on the polarization of an initial - state fermion in weak interactions .
the fermion lifetime was usually calculated in terms of spin states . however , after comparing spin states with helicity states and chirality states , it is pointed out that a spin state is helicity degenerate , and the spin state and the helicity state are entirely different . using helicity states ,
we calculate the lifetime of polarized muons .
the result shows that the lifetime of right - handed polarized muons is always greater than that of left - handed polarized muons with the same speed in flight . |
in this work , we present a novel approach to obtain a synthetic color - magnitude diagram ( cmd ) of galaxies , starting from predictions of chemical evolution models . our new _ photo - chemical model _ ` lights up ' the stars of chemical evolution models , according to their initial mass , metallicity and age ; in this way , we can obtain a new set of observational constraints for chemical evolution models beyond the usual photospheric chemical abundances . the method presented in this work can provide in principle a deeper insight in the interpretation of current observations , since we can understand how our hypothesis about galaxy formation and evolution can affect the final configuration of the cmd . by solving a set of physically - motivated differential equations , which take into account the main physical processes taking place and influencing the evolution of the galaxy interstellar medium ( ism ) ,
numerical codes of chemical evolution are able to provide the galaxy star formation history ( sfh ) and age - metallicity relation ; the evolution of the galaxy stellar and gas mass , and the run of the ism chemical abundances with time .
building up a photo - chemical code consists then in coupling the output of chemical evolution models with a database of stellar isochrones , currently available and computed with very high accuracy . -
and @xmath0-dimensions being @xmath1 ; the color coding in the figure corresponds to the number of stars within each grid element .
we consider in our analysis only clean isolated stellar detections and the sculptor member stars with an elliptical radius @xmath2 .
, width=351 ] most of the previous works in the literature recover the sfh of galaxies from the observed cmd by adopting sophisticated fitting techniques ( e.g. @xcite ) ; in particular , they search for the suitable linear combination of simple stellar populations ( ssps ) with different age and metallicity , which provides the best agreement with the observed photometric properties of the galaxy composite stellar population . as a byproduct
, this ` classical ' procedure can also predict an average galaxy age - metallicity relation .
nevertheless , no underlying approximate physical model is assumed in these works for the galaxy formation and evolution . in this paper ,
the first of a series of future works , we focus on reproducing the cmd of the sculptor dwarf spheroidal galaxy ( dsph ) .
in particular , we investigate whether the best chemical evolution model for sculptor reproducing the galaxy stellar metallicity distribution function ( mdf ) is able to predict a synthetic cmd which agrees with the observed one .
this work is organized as follows : in section [ sec : data ] we summarize the main properties of the sculptor dsph and describe the observed set of data used in this work for the comparison with our models ; in section [ sec : model ] we present the main characteristics of our photo - chemical model and the methods we employ to fairly compare the synthetic with the observed cmd ; in section [ sec : results ] we show our results , and in section [ sec : conclusion ] we draw some conclusions .
the sculptor dwarf galaxy was discovered by @xcite .
although it might appear simple at first glance , from the study of the kinematical , chemical and spatial distribution of its stellar populations , @xcite were able to disentangle in this galaxy an inner , kinematically ` cold ' , metal - rich stellar population from an outer ` hot ' metal - poor one , later on confirmed by @xcite and @xcite .
other studies based on photometric datasets also were able to identify ( or confirm ) the existence of stellar populations distinct in metallicity @xcite , age and spatial distribution @xcite .
@xcite reported for sculptor an average v - band surface brightness @xmath3 , an half - light radius @xmath4 , and an absolute v - band magnitude @xmath5 .
we make use of the distance modulus @xmath6 derived by @xcite . the observed cmd is taken from (
* see fig .
[ fig : cmd_observed ] ) , which were able to resolve stars down to the oldest main sequence turn - off ( msto ) of the sculptor dsph , by taking advantage of the deep wide - field photometry of ctio / mosaic . in order to avoid a non - negligible contamination of foreground mw disc field stars
, we consider only stars along the line - of - sight to the sculptor dsph with elliptical radius @xmath2 and corresponding to clean isolated detections ; the percentage of stars with these characteristics in the @xcite catalog is about @xmath7 per cent of the entire sample .
we find that , for @xmath8 , the noisy pattern introduced by foreground stars becomes larger than the `` signal '' of the sculptor cmd one wants to recover ( see also figure 5 in @xcite ) .
the observed stellar mdf is taken from @xcite , which combined the @xcite spectroscopic sample ( with sculptor stars belonging to the inner @xmath9 of elliptical radius ) with the dataset provided by the dwarf galaxies abundances and radial velocities team ( dart , covering a much larger radial extent and making use of the calcium triplet equivalent width to infer the fe abundances ; see @xcite ) , in order to have an mdf which were representative of the global sculptor stellar populations . by looking at (
* their figure a1 ) , the low - metallicity portion of their mdf is almost solely determined by the dart sample , with the @xcite mdf mainly contributing towards larger [ fe / h ] abundances .
, wind efficiency @xmath10 , infall mass @xmath11 , and infall time - scale @xmath12 .
the predicted mdf is convolved with a gaussian function with @xmath13 . ,
width=351 ] we make use of the parsec stellar isochrones @xcite , as computed for the following grid of stellar ages and metallicities , by assuming a reimers mass loss with efficiency @xmath14
the step in metallicity in our isochrone database is @xmath15 , from a minimum metallicity @xmath16 to a maximum metallicity @xmath17 .
the step in age between two adjacent isochrones is @xmath18 , from a minimum age @xmath19 to a maximum age @xmath20 . for self - consistency
, we assume in our model the same stellar lifetimes as the ones which can be derived from the parsec database ; in particular , we fit the stellar lifetimes with the following function : @xmath21},\ ] ] where @xmath22 , @xmath23 and @xmath24 are the fitting parameters , provided with the corresponding @xmath25-@xmath26 errors in the supplementary material , as functions of the metallicity @xmath27 . in fig .
[ fig : lifetimes ] we compare our derived stellar lifetimes with the ones of @xcite , which do not depend on metallicity and are extensively used in chemical evolution models .
the numerical code of chemical evolution is the same as the one adopted in @xcite where we address the reader for details for the study of the classical and ultra - faint dsph galaxies .
we make use of an updated version , by assuming the stellar yield compilation of ( * ? ? ?
* their model 15 ) and the stellar lifetimes derived from the parsec isochrones .
we assume the galaxy to assemble by accreting pristine gas from an external reservoir , until an infall mass given by @xmath28 is accumulated at @xmath29 .
the infall rate is assumed to follow a decaying exponential law , with typical time - scale @xmath30 .
we assume for the star formation rate the schmidt - kennicutt law , namely @xmath31 , where @xmath32 is the so - called star formation efficiency ( sfe ) and @xmath33 is the galaxy gas mass .
the run of the intensity of the sfr with time is crucially regulated by the various physical processes acting on @xmath33 , namely inflows and outflows of gas , returned matter from dying stars and supernovae , astration due to the star formation activity itself . a fundamental role in the evolution of dsphs
is played by the galactic outflows , which are predicted to occur very soon in these galaxies because of their shallow potential well ; the intensity of the outflow rate is assumed to be directly proportional to the sfr . on the one hand , if the efficiency for the gas removal is high ( typically @xmath34 ) , then the galaxy gas mass suddenly decreases and hence the sfr rapidly drops to zero ; on the other hand , if the galactic wind has a relatively lower efficiency ( typically @xmath35 ) , then the decrease in the sfr is smoother and it drops to zero on longer typical time - scales . per cent of the stars
observable at the present time are formed within the first @xmath36 of the galaxy evolution ; this time corresponds to the vertical dashed blue line in the figures .
furthermore , the number of stars with initial metallicity @xmath37 is roughly @xmath38 per cent of the total number of stars alive at the present time .
, width=351 ] we have explored the parameter space , by running a large number of chemical evolution models .
the best parameters for sculptor are found by minimizing the @xmath39 figure of merit , with the best model being the one reproducing the shape of the observed stellar mdf , which represents the most reliable observational constraint to any slight variation of the free parameters .
we vary the sfe in the range @xmath40-@xmath41 , the wind efficiency in the range @xmath42-@xmath43 , and the infall time - scale in the range @xmath44-@xmath45 . in order to allow our photo - chemical model to be more flexible when comparing its predictions with data
, we make the further approximation of assuming the galactic wind to be always active over the whole galaxy lifetime ; since all the galaxy physical properties in our chemical evolution model are normalized with respect to the infall mass , this assumption causes the predicted stellar mdf not to be influenced by the variation of the infall mass . in this simplified formulation of the galactic wind , all the galaxy physical properties ( such as the sfr , stellar mass , gas mass , and so on ) simply scale with the infall mass and our results can be easily readjusted for a different assumption of the cutoff in the elliptical radius of sculptor stars ( here we consider only stars with @xmath46 ) .
we find that the best chemical evolution model is characterized by the following parameters : 1 .
star formation efficiency @xmath47 ; 2 .
wind efficiency @xmath10 ; 3 .
infall time - scale @xmath12 .
we assume at the beginning a reference infall mass @xmath48 , as in @xcite , for which we predict a present - day total stellar mass @xmath49 .
the infall mass of the best model is then obtained by rescaling our results for the reference model so as to have the same number of stars in the synthetic and observed cmd .
we find for our best model an infall mass @xmath50 , giving rise to a present - day total stellar @xmath51 , larger than the value @xmath52 estimated by @xcite , but of the same order of magnitude .
according to the fitting formula of @xcite , which is assumed in many recent works to mimic a cosmologically motivated infall in galaxy formation and evolution models , dwarf galaxies with @xmath53 must have accreted almost @xmath54 per cent of their cumulative infall mass ( which turns out to be @xmath55 ) in the first @xmath56 of their evolution , namely from redshift @xmath57 to redshift @xmath58 , a larger timescale than the one found by our best model ( @xmath59 ) .
nevertheless , we remark on the fact that the @xcite fitting formula is strictly valid only in the redshift range @xmath60 , namely only after the reionization epoch ; therefore , a comparison with the gas mass assembly history of our model is not straightforward . in fig .
[ fig : mdf ] the predicted stellar mdf of the best model ( convolved with a gaussian function with @xmath61 ) is compared with the observed one @xcite .
the assumed very low sfe causes the mdf to be peaked at low @xmath62 $ ] abundances .
the width of the mdf is mostly determined by the wind efficiency ; in particular , the lower the @xmath63 parameter , the wider is the bulk of the galaxy star formation activity and hence also the mdf . in figure
[ fig : sfh]a ) we show the predicted sfh of our best model , while in figure [ fig : sfh]b ) we show the corresponding age - metallicity relation . in summary , the length of the bulk of the galaxy star formation activity can be regulated in our model by suitably varying the main parameters determining the star formation and outflow intensity and the galaxy gas accretion rate ; these parameters are the sfe , which determines the intensity of the sfr and the rate of thermal energy injection by supernovae , the wind efficiency , which determines the slope with which the sfr drops to zero , and the infall time - scale , which crucially determines the evolution with time of the galaxy potential well .
the stepwise structure of the photo - chemical model is the following . 1 .
we sample the galaxy sfh , as predicted by our best chemical evolution model for sculptor , to randomly extract an age for the formation of a given star .
2 . we sample the assumed initial mass function ( imf ) to randomly assign a mass to the star . in this work
we assume the @xcite imf for simplicity .
we use the age - metallicity relation of our best sculptor chemical evolution model to find the initial metallicity of the star .
4 . given the age , mass and metallicity of the star
, we check whether the star is alive or not at the present time , by assuming the metallicity dependent stellar lifetimes we have derived from the parsec stellar evolutionary tracks ( see section [ sec : lifetimes ] ) .
if the star can be observed at the present time , we store the photometric properties of the synthetic star , to later draw it in the synthetic cmd . on the one hand ,
the strength of our method resides in the very fine grid of the assumed isochrone database ; moreover , in our approach , we start from the predictions of chemical evolution models , assuming _ ab initio _ an underlying galaxy formation and evolution scenario , which is physically - motivated . on the other hand ,
the main shortcoming of our model is due to the fact that the lowest available metallicity in the parsec database is @xmath16 .
we assume that all the stars with @xmath64 have the same photometric properties as the stars with @xmath65 .
this fact can introduce a systematic error . by looking at figs .
[ fig : sfh]a ) and b ) , according to our best model , the galaxy spends its first @xmath66 at metallicity @xmath37 ; the number of stars with initial metallicity @xmath37 is roughly @xmath38 per cent of the total number of stars alive at the present time . to get a fair comparison with the observed cmd , we convolve the synthetic cmd with the distribution of the observed photometric errors , by assuming that the latter are gaussian .
in particular , we first divide the observed cmd in an uniform grid and , for each grid element @xmath67 , we compute the average v- and i - band observed photometric errors , @xmath68 , which we adopt as the standard deviations of the photometric noise in the @xmath67-th grid element . hence , for any given @xmath69-th synthetic star residing in the @xmath67-th grid element , we add the following noise to its predicted v- and i - band magnitudes : @xmath70 , where @xmath71 is a random number , drawn according to the standard normal distribution . in this way
, the model ` spreads out ' according to the errors in the data and we can fairly compare the synthetic with the observed cmd .
the synthetic cmd corrected for the photometric noise is then convolved with the results of the artificial star test performed by @xcite . in particular , by following a standard procedure , @xcite inserted in the observed images a large catalog of artificial stars ; after reducing and analyzing the altered images , they could compute the fraction of artificial stars recovered in the data as a function of their input magnitude and color .
we use the results of their test to compute the recovered fraction in each grid element , so as to throw out from the synthetic cmd the remaining lost fraction . by means of this type of analysis
, one can apply to the synthetic cmd the same completeness profile as is present in the data .
after correcting the synthetic cmd for the incompleteness , the number of synthetic stars strongly reduces , becoming @xmath72 .
in this section , we present the results of our photo - chemical model for the cmd of the sculptor dsph .
the main result of our work in shown in fig .
[ fig : cmd_result ] , where the synthetic cmd of sculptor ( left diagram ) is compared with the observed one ( right diagram ) . in order to better understand where the discrepancies between the observed and the synthetic cmd reside , in fig .
[ fig : residuals ] we plot the residuals , which correspond to the color - coding in the figure .
in particular , to better visualize the differences , we define the residual in the @xmath67-th grid element as : @xmath73 with @xmath74 and @xmath75 being the number of stars in the observed and synthetic cmd , respectively .
the regions of the observed sculptor cmd without any synthetic star are shown in fig .
[ fig : residuals ] as a greyscale density plot . on the one hand , by a visual inspection of figs .
[ fig : cmd_result ] and [ fig : residuals ] , we can obtain a quite good agreement for the red giant branch ( rgb ) , the horizontal branch ( hb ) and the asymptotic giant branch ( agb ) of the observed cmd . on the other hand , at fainter magnitudes , particularly in the sub - giant branch ( sgb ) and at the msto , the observed cmd extends towards slightly bluer colors than the synthetic one . moreover , our model can not reproduce the observed population of blue straggler stars which extend the main sequence towards blue colors and could be in principle reproduced by including the effect of merging binary stellar systems . we do not include blue straggler stars in our photo - chemical model .
we remark on the fact that it is not obvious that a model reproducing the chemical evolution of sculptor can also capture the main features of the observed galaxy cmd .
in fact , the final configuration of the synthetic cmd turns out to be highly affected by the variation of the free parameters of chemical evolution models . in particular , by increasing the sfe , the stellar metallicities accordingly increase at any fixed galactic time , causing the entire synthetic cmd to shift towards redder colors .
the imf acts in a similar way as the sfe , with the additional effect of filling up the various stellar evolutionary phases in the cmd with different relative fractions .
finally , the wind parameter and the infall time - scale crucially affect the spread of the predicted cmd , since they determine the length of the bulk of the galaxy star formation activity . in fig .
[ fig : vi_histo]a ) , we compare the predicted stellar @xmath76-color distribution ( black solid line ) with the observed one ( blue histogram ) ; this quantity turns out to be particularly sensitive to metallicity variations among the galaxy stellar populations . in fig .
[ fig : vi_histo]b ) , the predicted stellar luminosity function in the @xmath77-band ( black solid line ) is compared with the observed one ( blue line with error bars , which are shown as a shaded blue area ) ; the trend of this second quantity is more affected by the galaxy sfh and stellar lifetimes . an age indicator for the galaxy is given by the fraction of stars on the hb relative to the one on the rgb ; we predict @xmath78 . ) for the comparison between the observed and the synthetic cmd .
the greyscale density plot represents the regions of the observed sculptor cmd where no synthetic stars are predicted .
we find a good agreement for the rgb , hb and agb stars , whereas the msto and the sgb of the observed cmd extend towards bluer colors than the synthetic cmd .
we can not reproduce the population of blue straggler stars in the observed cmd , since we do not include them in our model .
, width=351 ] -color distribution ( black solid line ) with the observed one ( blue histogram with errorbars ) . in the bottom panel ,
we compare the predicted stellar luminosity function in the @xmath77-band ( back solid line ) with the observed one ( blue line , with the shaded blue area representing the @xmath25-@xmath26 errors ) .
msto stars are predicted to reside in the synthetic cmd at @xmath79 , whereas sgb stars at @xmath80 .
the peak in the @xmath77-band stellar luminosity function at @xmath81 is caused by hb stars .
, width=351 ] at a first glance , our photo - chemical model is able to predict and qualitatively reproduce the main features of the observed @xmath76-color distribution ; in particular , the first peak in fig .
[ fig : vi_histo]a ) ( the one at bluer colors ) is determined by the msto and sgb stars , whereas the second peak ( the one at redder colors ) is the signature left by the ascending rgb and hb stars . concerning the blue portion of the color distribution , from a visual inspection of fig . [
fig : vi_histo]a ) , we can not reproduce the observed population of blue straggler stars , which as aforementioned are not included in our photo - chemical model .
furthermore , a remarkable discrepancy resides in the decaying trend of the blue wing of the predicted color distribution , which contains a lower number of stars than the data , and in the predicted ` saddle ' , which turns out to be higher than the observed one .
this can be also appreciated by looking at the residual plot in fig [ fig : residuals ] , where the observed cmd clearly contains a larger number of msto and sgb stars with blue colors than the synthetic cmd .
this discrepancy seems likely the signature of metal - poor stellar populations in the sculptor dsph , which our one - zone chemical evolution model has not been able to capture . nevertheless , in principle
, it could also indicate a predicted age - metallicity relation which is steeper than what seems to be required by observations ; in fact , one would obtain a similar discrepancy if also the synthetic metal - rich stars are older than the observed ones .
interestingly , by looking at fig .
[ fig : mdf ] , the observed stellar mdf suggests the presence of two distinct peaks , corresponding to two separated main stellar populations in the galaxy .
the latter feature can not be resolved by our best chemical evolution model , which indeed predicts the stellar mdf to have a single peak , lying between the two of the observed distribution .
we remark on the fact that this feature is peculiar to the @xcite mdf , which combines the dart sample determining the low metallicity portion of the observed mdf with the one by @xcite , concentrated towards slightly higher metallicity .
the regions where the observed cmd contains a larger number of blue ( metal - poor ) stars than the synthetic one likely correspond to the low - metallicity peak in the observed mdf , which is also the most pronounced one .
accordingly , the higher ` saddle ' in the predicted color distribution ( see the top panel in fig . [
fig : vi_histo ] ) confirms that the model predicts galaxy stellar populations which are intermediate between the observed metal - poor and metal - rich ones .
the observed stellar mdf , as derived by @xcite , includes stars with @xmath82 , namely beyond the radial cutoff we assume for the observed cmd .
as one moves towards the outer zones of the galaxy , the sculptor stellar populations are observed to become increasingly old and metal - poor , determining the observed stellar mdf at low metallicity .
nevertheless , the relative number of stars also increasingly diminishes when moving outwards , with almost @xmath83 per cent of the @xcite sample being contributed by stars with @xmath84 .
furthermore , the shape of the observed stellar mdf in the inner radial bins does not significantly vary at low metallicity with respect to the mdf in the outer bins ( see figure 14 of @xcite ) . therefore ,
if we had applied a radial cutoff at @xmath85 also to the observed mdf , our analysis would not have been significantly altered . a viable solution to reduce the discrepancy between the observed and the synthetic cmd would be , for example , to disentangle the two different stellar populations in the observed mdf , by assuming two underlying separated distributions , peaked at slightly lower and higher @xmath62 $ ] abundances than the predicted mdf of the best model . in this way
, the metal - poor stellar population could be reproduced by assuming a lower sfe than the metal - rich one .
after superimposing the two stellar populations in the synthetic cmd with appropriate weights , one would extend the synthetic cmd towards slightly bluer color , hence obtaining a better agreement with the observed cmd .
finally , we can not exclude that the lack of binary stars in our photo - chemical model might also contribute to the discrepancy between the observed and the synthetic cmd , since their inclusion would cause a broadening of the ms and therefore a redistribution of the colors of the synthetic stars in fig .
[ fig : vi_histo]a ) . in fig .
[ fig : vi_histo]b ) , the observed stellar @xmath77-band luminosity function is compared with the predicted one . for the rgb and agb stars
, there is a good agreement between the model and data .
furthermore , the model predicts a peak at @xmath86 , which represents the effect of hb stars ; the presence of this peak is not visible in the observed @xmath77-band stellar luminosity function because of the large foreground contamination in the redder part of the observed cmd , both at fainter and at brighter @xmath77-band magnitudes than the ones of the observed hb . if we had considered only sculptor stars in the innermost regions of the galaxy ( e.g. with @xmath87 ) , we would have obtained a well agreement between model and data also for the hb stars . at fainter magnitudes , the model in fig .
[ fig : vi_histo]b ) contains a larger number of msto and a lower number of sgb stars than the data ; this discrepancy might be partly due to the inherent uncertainty in the assumed completeness profile , which is particularly important for these evolutionary stages , as well as it could also indicate the need for an imf with a lower number of low - mass stars than the @xcite imf assumed in this work .
in this work , we have presented a new approach to draw the synthetic cmd of galaxies .
in particular , we have developed _ ab - initio _ a new photo - chemical model , which we have applied to reproduce the observed cmd of sculptor dsph .
our numerical code starts from the predictions of chemical evolution models about the galaxy sfh and age - metallicity relation .
then , by assuming the parsec stellar evolutionary tracks , we can ` light up ' the stars with different age , mass and metallicity of chemical evolution models , in order to draw a synthetic cmd .
we have defined the best chemical evolution model for sculptor as the one reproducing the observed galaxy stellar mdf .
several improvements could be done in our photo - chemical model , by considering for example an underlying cosmological framework , whose primary effect would be to influence the evolution of the galaxy gas mass assembly with time .
interestingly , a very first attempt to draw the cmd of a dsph galaxy within a full cosmological framework by making use of a semi - analytical model for the galaxy formation and evolution is represented by the work of @xcite , which adopted the freely available iac - star code @xcite , however , they did not provide a detailed discussion of their findings about the predicted galaxy cmd .
a further improvement in our photo - chemical model would be to include the effects of unresolved binary stars and foreground contamination in the synthetic cmd .
the strength of our approach resides in the statistical sampling of the galaxy predicted sfh and assumed imf , as well as in the assumption of a very fine grid of stellar isochrones , both in metallicity and in age .
the main shortcoming is related to the parsec stellar isochrones , which are computed only for @xmath88 .
the sfh can be regulated in our models for dsphs by suitably varying the main parameters triggering the onset of the galactic wind and determining its intensity .
in fact , once the galactic wind has started , the sfr rapidly drops to zero , since most of the infall mass has been accumulated fast within short typical time - scales .
our best model for sculptor predicts that @xmath89 per cent of the stars observable at the present time are formed within the first @xmath36 of the galaxy evolution .
we predict at the present time a total stellar mass @xmath51 , which is of the same order of magnitude as other estimates like @xmath52 by @xcite .
also the predicted evolution of the sfr as a function of time is in agreement with the findings of @xcite .
stellar systems or interstellar regions with low gas density , such as low - mass dwarf galaxies or the outer parts of spiral galaxies , likely follow a star formation law which deviates from the usually assumed schmidt - kennicut law , @xmath31 ; for this reason , we have done some numerical experiments by assuming the same expression for the star formation rate as in the original work of @xcite ( see also @xcite for a detailed study in the context of hydrodynamical simulations ) . by assuming the @xcite law , we predict the sfh to be more concentrated in the earliest epoch of the galaxy evolution and the metallicity z to initially evolve more rapidly than our best - fitting model ; then , at later times , z remains quite constant when the @xcite law is assumed , while it increases in our best - fitting model .
finally , we find that the final total gas and stellar mass are almost the same when the two different expressions for the star formation rate are assumed .
we have shown that our photo - chemical model is able to capture the main features of the observed cmd of the sculptor dsph , with the best agreement being obtained for the rgb , hb and agb stars .
the discrepancy has been found at fainter luminosity in the msto and sgb , where the observed cmd extends towards bluer colors than the synthetic one . that may be caused by underlying metal - poor stellar populations which our photo - chemical model has not been able to capture as well as to the lack of binary stars in our model , which would also broaden the synthetic cmd at faint magnitudes .
in fact , the predicted stellar mdf is characterized by a single peak , whereas the observed one suggests the presence of two peaks , residing at slightly lower and higher @xmath62 $ ] abundances than the model peak . in particular , the more pronounced peak in the observed
mdf corresponds to the one at lower @xmath62 $ ] abundances .
therefore , our photo - chemical model contains stellar populations which are intermediate between the metal - poor and the metal - rich ones in the observed stellar mdf . in order to reduce the discrepancy
, one could superimpose the results of multiple one - zone chemical evolution models and find the linear combination which provides the best agreement with the observed stellar mdf
. this will be the subject of a future work , in which we will also show the effect of varying the free parameters of chemical evolution models on the synthetic cmd .
although the uncertainties in the assumed completeness profile can be important at the msto and sgb , the discrepancy between model and data in the @xmath77-band stellar luminosity function for @xmath90 might be alleviated by assuming an imf which contains a lower number of low - mass stars than the @xcite imf assumed in this work and by including the effect of binary stars .
fv thanks e. brocato for insightful discussions during the visit at the astronomical observatory of rome in 2014 december , and s. recchi for his precious suggestions .
fm and mt acknowledge financial support from prin - miur 2010 - 2011 project ` the chemical and dynamical evolution of the milky way and local group galaxies ' , prot .
2010ly5n2 t .
we thank an anonymous referee for his / her constructive comments . | we present a novel approach to draw the synthetic color - magnitude diagram of galaxies , which can provide in principle a deeper insight in the interpretation and understanding of current observations .
in particular , we ` light up ' the stars of chemical evolution models , according to their initial mass , metallicity and age , to eventually understand how the assumed underlying galaxy formation and evolution scenario affects the final configuration of the synthetic cmd . in this way
, we obtain a new set of observational constraints for chemical evolution models beyond the usual photospheric chemical abundances .
the strength of our method resides in the very fine grid of metallicities and ages of the assumed database of stellar isochrones . in this work
, we apply our photo - chemical model to reproduce the observed cmd of the sculptor dsph and find that we can reproduce the main features of the observed cmd .
the main discrepancies are found at fainter magnitudes in the main sequence turn - off and sub - giant branch , where the observed cmd extends towards bluer colors than the synthetic one ; we suggest that this is a signature of metal - poor stellar populations in the data , which can not be captured by our assumed one - zone chemical evolution model .
-.6 in [ firstpage ] local group galaxies : dwarf galaxies : stellar content stars : abundances hertzsprung - russell and colour - magnitude diagrams |
the qcd amplitudes have a complicated analytic structure in the complex plane of the coupling constant , @xmath1 .
as proved in @xcite , the infinitely many multiparticle branch points at large energies result , via renormalization group invariance , in an accumulation of essential singularities near the origin @xmath2 .
since in massless qcd the multiparticle hadronic states are generated only by a nonperturbative confinement mechanism , these singularities can show up only beyond perturbation theory .
however , the perturbative amplitudes themselves are expected to have a complicated structure of singularities , due to the fact that the perturbation series is divergent and borel non - summable .
the presence of the renormalons on the real axis of the borel plane induces singularities of the amplitudes as functions of the coupling constant . in a recent paper @xcite
, arguments based on analyticity in the coupling complex plane were used to suggest the possibility of calculating genuine nonperturbative quantities , like qcd condensates , from pure perturbation theory .
the analytic structure and its connection with infrared renormalons were further discussed in @xcite .
motivated by this recent interest in the problem , we investigate in the present work the analytic structure in the complex coupling plane of the borel - summed amplitudes in perturbative qcd .
we use the mathematical techniques applied in @xcite , which allow us to express the borel integral as a dispersion relation in the coupling plane . in the last section
we shall make a few comments on the papers @xcite .
we consider for illustration the adler function in massless qcd @xmath3 where @xmath4 is the current - current correlation function calculated for euclidian arguments @xmath5 .
it is known that the perturbation expansion of @xmath6 in powers of the renormalized coupling @xmath7 is divergent and not borel summable ( see @xcite and references therein ) .
the attempts of performing the summation by a formal borel - laplace integral @xcite encounter the difficulty that this integral is not well - defined .
we consider the borel transform @xmath8 defined in the standard way in terms of the perturbative coefficients @xmath9 of @xmath10 : @xmath11 where @xmath12 is the first qcd beta - function coefficient with @xmath13 the number of flavors .
it is known , from the @xmath14 large order growth of @xmath9 , that @xmath8 has singularities ( ultraviolet and infrared renormalons ) on the real axis of the @xmath15-plane @xcite . for the adler function , the ultraviolet renormalons are placed along the range @xmath16 and the infrared renormalons along @xmath17 ( see fig . [ fig1 ] ) . due to the infrared renormalons ,
the usual borel - laplace integral is not well - defined and requires an integration prescription . defining @xmath18
one can adopt as prescription , for each value of @xmath0 with @xmath19 , either @xmath20 or @xmath21 , or a linear combination of them , with coefficients @xmath22 and @xmath23 such as to correctly reproduce the known low - order expansion of @xmath6 ( which is obtained by truncating the taylor expansion ( [ b ] ) at a finite order @xmath24 ) .
we consider in particular the principal value ( @xmath25 ) prescription @xmath26\,.\ ] ] once a prescription is adopted , one has a well - defined function of @xmath0 , different prescriptions yielding different functions . in the next section we shall study the analytic properties of these functions in the complex @xmath0-plane .
the analytic properties of the integrals ( [ pm ] ) with respect to the variable @xmath0 can be studied with standard mathematical techniques . in the present work ,
we use the method of inverse mellin transform , applied for the first time in the context of borel summation in qcd in @xcite . for details of the mathematical procedure used below ,
we refer to @xcite , where the same method was applied for investigating the momentum - plane analyticity structure of the adler function in the large-@xmath27 limit .
the inverse mellin transform of the function @xmath28 is defined by @xcite @xmath29 and admits the inverse relation @xmath30 which gives @xmath31 in the strip @xmath32 parallel to the imaginary axis @xcite .
the above relations are valid if the borel transform satisfies the condition @xcite @xmath33 where @xmath34 is a point located on the real axis , between the branch points , @xmath35 ( see fig.1 ) .
this condition strongly restricts the asymptotic behaviour of the borel transform , and it is not known whether it is obeyed or not in qcd .
the condition ( [ l2b ] ) is however satisfied in some particular cases of physical interest .
one example is the summation of a chain of diagrams in the large-@xmath27 ( or large-@xmath13 limit ) , which leads to the borel transform @xcite : @xmath36 ^ 2 } \,,\ ] ] working in the @xmath37-scheme , where all the exponential factors are included in the definition of the coupling . as a second example
we take the case of a finite number of renormalons with branch - point singularities , in particular the contribution to @xmath8 of the leading infrared and ultraviolet renormalons @xmath38 where @xmath39 , @xmath40 and the constants @xmath41 and @xmath42 represent the strength of the corresponding singularities . we note that for the adler function @xmath43 , @xmath44 , and the exponents @xmath45 and @xmath46 were calculated in @xcite and @xcite , respectively .
the function @xmath47 defined in ( [ wd ] ) can be calculated by closing the integration contour along a semi - circle at infinity in the @xmath15 plane and applying the theorem of residues .
for @xmath48 the contribution of the semi - circle at infinity vanishes if the contour is closed in the right half of the @xmath15 plane , while for @xmath49 the contour must be closed in the left half plane .
therefore one obtains different expressions for the distribution function at @xmath48 and @xmath49 . by cauchy s theorem these functions , which we denote by @xmath50 and @xmath51 , have the representations @xmath52 the integration ( hankel ) contours @xmath53 and @xmath54 being indicated in fig . 1 .
the inverse mellin transform @xmath55 was calculated in ref .
@xcite for the adler function in the large-@xmath27 limit , when the borel transform has the expression ( [ blb0 ] ) . in this case
@xmath56 \right\ }
\,,\\ \widehat w_>(\tau)&= & \frac{32}{3 } \left\ { 1 + \ln\tau + \left ( \frac34 + \frac12 \ln\tau \right ) \frac{1}{\tau } + ( 1+\tau)\big [ l_2(-\tau^{-1 } ) - \ln\tau \ln(1+\tau^{-1 } ) \big ] \right\ } , \nonumber\end{aligned}\ ] ] where @xmath57 is the dilogarithm .
the physical interpretation of @xmath55 as the distribution of the internal gluon virtualities in feynman diagrams was also pointed out in @xcite .
the function @xmath55 can be also calculated explicitly for a finite number of renormalons with branch - point singularities .
for instance , for the borel transform written in ( [ bfirst ] ) we obtain @xcite : @xmath58 we now proceed to the evaluation of the integrals ( [ pm ] ) , taking @xmath0 in the right half plane , @xmath19 , where we assume that they converge .
our aim is to obtain a representation of @xmath59 in terms of the inverse mellin transform @xmath55 . to this end
we rotate the integration contours @xmath60 , whitout crossing singularities , up to a line parallel to the imaginary axis , where the representation ( [ wdinv ] ) is valid .
it is easy to check that for @xmath0 in the upper half plane ( @xmath61 ) , the contribution of the quarter of the circle at infinity vanishes if the rotation is performed in the upper half of the @xmath15-plane . using the representation
( [ wdinv ] ) , valid along the imaginary axis , and performing the integral with respect to @xmath15 ( for details see @xcite ) , we obtain @xmath62 similarly , for @xmath0 in the lower half plane ( @xmath63 ) the integration axis in the expression of @xmath21 can be rotated in the lower half of the @xmath15-plane , up to the negative imaginary axis , leading to @xmath64 we notice further , recalling the definition ( [ pm ] ) , that the first relation ( [ wd1 ] ) can be expressed as @xmath65 where we introduced the notation @xmath66_{\tau = e^{-{1\over \beta_0 a}}}\,.\ ] ] in particular , using the expression ( [ wfirst ] ) of @xmath55 , we obtain for the leading renormalons : @xmath67 it is important to emphasize that the relation ( [ dif ] ) is valid for @xmath19 ( or equivalently for @xmath48 ) , i.e. in the whole right - half of the @xmath0-plane .
the relations ( [ dp1 ] ) , ( [ dm1 ] ) and ( [ dif ] ) are the basis the derivation of the dispersion relations given below .
we first note that , unlike the original representations ( [ pm ] ) which converge only for @xmath68 , the representations ( [ dp1 ] ) and ( [ dm1 ] ) can be analytically continued in the corresponding upper ( lower ) half of the complex @xmath0 plane , outside the real axis . moreover , it is easy to convert them into a dispersion representation in the variable @xmath0 .
we first split the integral in two integrals , one from 0 to 1 ( where @xmath69 ) , and the other from 1 to @xmath70 ( where @xmath71 ) , and perform in each interval the change of variable @xmath72 using the relations ( [ dp1])-([dm1 ] ) thus transformed , together with ( [ dif ] ) and ( [ sigma ] ) , we finally express @xmath20 as @xmath73 from the definition ( [ sigma ] ) and the properties of the inverse mellin transform @xmath55 , it follows that the function @xmath74 can be analytically continued in the complex @xmath0-plane .
this property is seen explicitly in the case of one infrared renormalon in ( [ sigma1 ] ) .
thus , the expressions ( [ drap ] ) are analytic functions in the upper ( lower ) half of the complex @xmath0-plane , outside the real axis . from ( [ sigma1 ] ) it is seen that @xmath75 is real for @xmath76 , while @xmath74 is real for @xmath77 .
therefore , the spectral functions of the dispersion integrals are real . taken together , the dispersion relations ( [ drap ] ) define a single analytic function , @xmath20 , in the whole cut @xmath0-plane .
this function may have a discontinuity across the positive axis , due the cauchy dispersion integral along @xmath78 and the additional term @xmath79 in the second relation .
a simple calculation shows however that @xmath80 where p denotes the cauchy principal value .
this relation shows that the function @xmath20 is well defined along the positive axis , but has there an unphysical imaginary part equal to @xmath81 from ( [ drap ] ) it follows that @xmath20 has actually a discontinuity along the negative axis , given by @xmath82 \,,\quad a<0\,,\ ] ] in terms of the real spectral function @xmath83 and the analytic continuation of @xmath84 up to the lower edge of the negative semiaxis ( where , as seen from ( [ sigma1 ] ) , it is complex ) . the function @xmath21 satisfies a dispersion relation similar to ( [ drap ] ) , with an additional term @xmath85 in the right hand side , as follows from ( [ dif ] ) . as above
, one can show that @xmath21 is well - defined for @xmath77 , but it assumes there complex values .
the unphysical imaginary part is eliminated if we adopt the principal value prescription defined in ( [ pv ] ) , for which we obtain : @xmath86 for @xmath0 on the positive semiaxis , is easy to check that @xmath87 therefore , @xmath88 is well - defined and real on the positive real semiaxis , as required by general principles .
the expressions ( [ drppv ] ) define an analytic function which satisfies the reality condition @xmath89^*$ ] in the whole complex @xmath0-plane cut along the negative semiaxis , where it has a discontinuity @xmath90\,,\quad a<0.\ ] ] as argued in @xcite , this discontinuity vanishes only under strong restrictions on the asymptotic behavior of @xmath8 , which are satisfied neither in the simple cases considered here , nor , most probably , in full qcd : namely , the @xmath91 condition ( [ l2b ] ) must be satisfied not only by @xmath8 , but also by the product @xmath92 ( for technical details , see ref .
@xcite ) . we point out that the imaginary part of @xmath88 along the positive semiaxis vanished due to a precise cancellation of the imaginary part of the integrals and the last terms in eqs .
( [ drppv ] ) .
it is easy to check that for a general linear combination of @xmath93 , with coefficients @xmath22 and @xmath23 as discussed above ( [ pv ] ) , this cancellation no longer holds .
therefore , the principal value prescription is the most suitable choice if one wants to preserve in perturbative qcd the analytic properties of the true amplitudes . in this section
, we obtained the nontrivial result that the inverse mellin transform can be used to derive from ( [ pm ] ) dispersion relations for the perturbative amplitudes @xmath94 , which allow an analytic continuation of the borel integral into the left - hand half - plane @xmath95 and explicitly exhibit the singularities and the discontinuities across the cuts
. the results obtained here , besides expressing the borel integral ( [ pm ] ) in the more suitable form of a dispersion relation , will be useful in discussing the validity of some assumptions on analyticity in the @xmath0 plane made in the literature , as we show in the next section .
in this section we shall make a few comments on the recent papers @xcite , related to the present work . in ref .
@xcite the author considers , in the case of a single infrared renormalon , the problem of removing the unphysical imaginary part of @xmath20 by subtracting a suitable regularization function from it .
our results in this particular case are consistent with @xcite .
indeed , eqs .
( [ drap ] ) and ( [ drppv ] ) show that the function @xmath88 is obtained by subtracting from @xmath20 the function @xmath96 which , using ( [ sigma1 ] ) , can be written for @xmath97 as @xmath98 \,.\ ] ] this expression coincides with eq .
( 17 ) of @xcite , derived using arguments based on regularity with respect to the parameter @xmath45 . the analytic structure in the coupling constant plane was recently considered also in ref .
@xcite , where arguments based on analyticity were used in support of the claim that the qcd condensates can be determined using the coefficients of the perturbation series . in
what follows we shall briefly analyse the validity of this claim .
the author of @xcite uses the analogy with some semiclassical models @xcite , where a specific contribution ( for instance , multi - instantons ) is regular for negative couplings and can be obtained for positive couplings by analytic continuation .
he invokes the heuristic argument according to which a perturbation series with a sign - nonalternating , @xmath14 large order behavior , can be summed at @xmath76 by a borel integral along the negative axis in the @xmath15-plane , where it becomes sign alternating .
we note however that sign alternation does not necessarily imply borel summability : this property is violated if a nonalternating component is present , however negligible it may be in comparison with a strong sign - alternating component of the series .
this is exactly the situation in qcd : due to the ultraviolet renormalons , the borel integral along the negative @xmath15-axis is not well defined .
another conjecture adopted in @xcite is related to the nonperturbative amplitude @xmath99 , which must be added to @xmath6 in order to compensate its unphysical imaginary part : @xmath100 according to current interpretations @xcite , the cancellation is expected to occur if both terms in the above relation are calculated with the same prescription . in @xcite the author supplements ( [ cond ] ) by a specific assumption about the nonperturbative amplitude , taking it of the form @xmath101 where @xmath45 is the branch - point exponent of the infrared renormalon in ( [ bfirst ] ) and @xmath102 is related to the gluon condensate ( we use the parametrization given in @xcite , with our notation @xmath103 ) .
this expression is regular for negative @xmath0 , having a discontinuity _ only _ on the positive semiaxis in the @xmath0-plane , where @xmath104 \ , , \quad a>0\,.\ ] ] using this relation and the imaginary part of the perturbative amplitude given in ( [ i m ] ) and ( [ sigma1 ] ) , condition ( [ cond ] ) gives the relation @xmath105 from which , according to @xcite , one could obtain the nonperturbative parameter @xmath102 ( the gluon condensate ) , using the strength @xmath41 of the infrared renormalon computed from the perturbation series .
the relation ( [ residue ] ) implies that @xmath41 vanishes for nonnegative integer @xmath45 , when the renormalon @xmath106 becomes a pole .
it is known however that poles are actually obtained from some chains of feynman diagrams , in the large-@xmath27 ( or large-@xmath13 ) limit @xcite . in @xcite
the author discusses this limit , taking the exponent @xmath45 of the form @xmath107 , with @xmath108 an integer . in this limit ,
the factor @xmath109 in the denominator of ( [ residue ] ) is compensated by the factor @xmath27 in the numerator , leading to a finite nonzero limit for @xmath41 , if @xmath102 tends to a nonzero constant in the large-@xmath27 limit .
but then eq . ( [ dnp ] ) implies that the real part of the nonperturbative amplitude @xmath110 is nonvanishing in this limit ( we keep , following @xcite , the product @xmath111 constant , which is legitimate , as seen in particular in the one - loop expression @xmath112 ) . on the other hand ,
the perturbative amplitude @xmath113 is subleading for large @xmath27 , due to the factor @xmath114 in front of the laplace - borel integral ( [ pm ] ) .
a choice of the constant @xmath102 of the same order , i.e. @xmath115 , would imply from ( [ residue ] ) that @xmath41 vanishes when @xmath116 .
so , it follows from @xcite that in the large-@xmath117 limit either the total adler function is dominated by the real part of the nonperturbative amplitude , or the renormalon residue vanishes .
the above implications depend however , in a crucial way , on the conjecture made in @xcite that the only singularities of the nonperturbative amplitude are along the positive semiaxis .
actually , singularities along the negative semiaxis ( produced by the ultraviolet renormalons in the ope coefficients ) can not be excluded . in @xcite
it is claimed that the influence of the ultraviolet renormalons can be suppressed by an appropriate conformal mapping @xcite . a numerical suppression , however
, does not imply removal of the corresponding singularities . as is shown in @xcite , even if @xmath8 is expanded in powers of the optimal conformal mapping of the borel plane @xcite
, the function @xmath6 does have singularities along the negative semiaxis .
therefore , the general expression of @xmath99 may have singularities both for @xmath77 and for @xmath76 . to give an example ,
we add to ( [ dnp ] ) the analogous branch - point term suggested by eq.([delta ] ) , taking @xmath118 where @xmath102 and @xmath119 depend in general on @xmath45 and @xmath120 is complex ( as the condition ( [ cond ] ) holds in a definite prescription , we specifically refer here to the prescription leading to @xmath121 )
. then condition ( [ cond ] ) gives @xmath122 which does not necessarily imply that @xmath41 vanishes when @xmath45 is a nonnegative integer .
moreover , the real part of the nonperturbative amplitude ( [ dnp1 ] ) contains the additional parameter @xmath123 not specified by the condition ( [ cond1 ] ) , which , combined with the first term in the expression ( [ dnp1 ] ) can make @xmath124 subleading in the large-@xmath27 limit .
this counterexample disproves the conclusions ( that appear to be inherent in the method of @xcite ) that the renormalon residue vanishes when the singularity is a pole , or the amplitude is dominated in the large-@xmath27 limit by the nonperturbative part .
even more important , however , is the fact that the real part of the nonperturbative amplitude can not be determined from the requirement ( [ cond ] ) and analyticity arguments .
the possibility , advocated in @xcite , of obtaining the gluon condensate from the perturbation series fails to work here . in the present paper , by using the inverse mellin transform , we showed that the borel - summed qcd perturbative amplitudes ( [ pm ] ) satisfy dispersion relations which explicitly exhibit the singularities in the complex @xmath0-plane and the discontinuities across the cuts , and also allow the analytic continuation of the borel integral into the left - hand half - plane @xmath95 .
we saw that the derivation relies on some special hypotheses about the properties of the borel transform in perturbative qcd : we namely assumed that the borel transform has no singularities in the complex @xmath15-plane except for branch - points on the real axis , with a holomorphy gap around the origin , its asymptotic behavior being such that the inequality ( [ l2b ] ) holds and the inverse mellin transform ( [ wd ] ) exists .
we point out that the holomorphy of @xmath8 in the double - cut @xmath15-plane is expected on general grounds to hold in renormalisable field theories @xcite .
the validity of this condition in qcd is presently almost universally adopted as a plausible assumption . as for the asymptotic condition ,
its validity in full qcd is not guaranteed , but it is satisfied in the large-@xmath27 limit and for a finite number of renormalons with branch - point singularities . in the second part of the paper we made some comments on the papers @xcite , where arguments based on analyticity were used to discuss the connection between renormalons and nonperturbative quantities in qcd .
we investigated some specific conjectures made in @xcite and showed that they are in conflict with the analyticity properties related to the ultraviolet and infrared renormalons , which throws doubts on the method of calculating qcd condensates from perturbative expansions .
we thank prof .
jiri chyla for many interesting and stimulating discussions on topics related to this work .
we also thank dr .
t. lee for useful correspondence .
one of us ( i.c . ) thanks prof .
s. randjbar - daemi and prof . c. verzegnassi for hospitality at `` abdus salam '' ictp and infn sezione di trieste .
this work was supported by the ceres program of romanian mect , by the romanian academy under grant nr .
23/2003 , and by the ministry of industry and trade of the czech republic , project rp-4210/69 .
99 g. t hooft , in : _ the why s of subnuclear physics _ , proceedings of the 15th international school on subnuclear physics , erice , sicily , 1977 , edited by a. zichichi ( plenum press , new york , 1979 ) , p. 943 .
n. n. khuri , phys .
d*23 * , 2285 ( 1981 ) .
t. lee , phys .
rev . d*66 * , 034027 ( 2002 ) .
g. cveti , phys .
rev . d*67 * , 074022 ( 2003 ) .
t. lee , hep - ph/0303240 .
i. caprini and m. neubert , jhep * 03 * , 007 ( 1999 )
. i. caprini and j. fischer , phys.rev . * d62 * , 054007 ( 2000 ) eur.phys .
j. c*24 * , 126 ( 2002 ) .
m. beneke , phys . rep . * 317 * , 1 ( 1999 ) .
g. g. hardy , _ divergent series _ , oxford univ .
press , new york , 1949 ) ; a. sokal , j. math . phys .
21 , 261 ( 1980 ) ; j. fischer , int.j .
a*12 * , 3265 ( 1997 ) .
m. neubert , phys .
d*51 * , 5924 ( 1995 ) .
p. ball , m. beneke and v.m .
braun , nucl phys .
b*452 * , 563 ( 1995 ) .
titchmarsh , _ introduction to the theory of fourier integrals _ , clarendon press .
oxford , 1948 .
m. beneke , phys .
b * 307 * , 154 ( 1993 ) ; nucl .
b * 405 * , 424 ( 1993 ) .
d. broadhurst , z. physik c * 58 * , 339 ( 1993 ) .
mueller , in _ qcd - twenty years later _ , aachen 1992 , edited by p. zerwas and h. a. kastrup , world scientific , singapore , 1992 .
m. beneke , v.m .
braun and n. kivel , phys .
b * 404 * , 315 ( 1997 ) .
a. erdlyi ed .
, _ tables of integral transforms _
, bateman manuscript project , vol .
i , mcgraw - hill , new york 1954 .
j. zinn - justin , _ quantum field theory and critical phenomena _ , oxford univ
. press , 1996 i. caprini and j. fischer , phys .
rev . d*60 * , 054014 ( 1999 ) .
g. parisi , nucl .
b*150 * , 163 ( 1979 ) .
mueller , nucl.phys .
b * 250 * , 327 ( 1985 ) . | we investigate the analytic structure of the borel - summed perturbative qcd amplitudes in the complex plane of the coupling constant . using the method of inverse mellin transform ,
we show that the prescription dependent borel - laplace integral can be cast , under some conditions , into the form of a dispersion relation in the @xmath0-plane .
we also discuss some recent works relating resummation prescriptions , renormalons and nonperturbative effects , and show that a method proposed recently for obtaining qcd nonperturbative condensates from perturbation theory is based on special assumptions about the analytic structure in the coupling plane that are not valid in qcd . |
the solar convection zone is highly turbulent and mixing is expected to be efficient .
nevertheless , the sun displays coherent structures encompassing many turbulent eddy scales .
a well - known example is the large - scale magnetic field of the sun that is antisymmetric about the equator and shows a 22 year solar cycle @xcite .
another prominent example in the sun is the emergence of active regions .
it is generally believed that active regions are the result of some non - axisymmetric instability of @xmath0 magnetic fields in the tachocline @xcite .
however , the existence of such strong fields remains debatable @xcite . a powerful tool for understanding the emergence of such large - scale structures from a turbulent background is mean - field dynamo theory @xcite . with the advent of powerful computers and numerical simulation tools , it has become possible to confront many of the mean - field predictions with direct simulations @xcite .
here we consider the idea that statistically steady , stratified , hydromagnetic turbulence with an initially uniform magnetic field is unstable to the negative effective magnetic pressure instability ( nempi ) .
this instability is caused by the suppression of turbulent hydromagnetic pressure ( the isotropic part of combined reynolds and maxwell stresses ) by the mean magnetic field @xcite . at large reynolds numbers and for sub - equipartition magnetic fields ,
the negative turbulent contribution can become so large that the effective mean magnetic pressure ( the sum of turbulent and non - turbulent contributions ) appears negative . in a stratified medium , this results in the excitation of nempi that causes formation of large - scale inhomogeneous magnetic structures .
nempi is similar to the large - scale dynamo instability , except that it only redistributes the total magnetic flux , creating large - scale concentrated magnetic flux regions at the expense of turbulent kinetic energy .
historically , the magnetic suppression of the reynolds stress was first found by @xcite and @xcite .
later , @xcite considered the maxwell stress and found the mean effective magnetic tension to be suppressed by mean fields . however , these calculations were based on quasi - linear theory which is only valid at low fluid and magnetic reynolds numbers . @xcite considered the combined reynolds and maxwell stresses at large reynolds numbers and found a sign reversal of the effective mean magnetic pressure .
this result is based on the @xmath1 approximation , and has been corroborated using the renormalization procedure @xcite .
the magnetic suppression of the combined reynolds and maxwell stresses is quantified in terms of new turbulent mean - field coefficients that relate the components of the sum of reynolds and maxwell stresses to the mean magnetic field .
these coefficients depend on the magnetic field and have now been determined in direct numerical simulations ( dns ) for a broad range of different cases , including unstratified forced turbulence @xcite , isothermally stratified forced turbulence ( * ? ? ?
* hereafter bkkr ) , and turbulent convection @xcite .
these simulations have clearly demonstrated that the mean effective magnetic pressure is negative for magnetic field strengths below about half the equipartition field strength .
however , these dns studies did not find the actual instability . with a quantitative parameterization in place , it became possible to build mean - field models of stratified turbulence which clearly demonstrate exponential growth and saturation of nempi . in view of applications to the formation of active regions in the sun ,
such simulations were originally done for an adiabatically stratified layer @xcite .
in addition , mean - field studies showed the existence of nempi even for isothermal stably stratified layers @xcite .
this last result turned out to be important because it paved the way for this letter where we demonstrate nempi through dns .
once we establish the physical reality of this effect , it would be important to apply it to realistic solar models which include proper boundary conditions , realistic stratification , convective flux , and radiation transport .
however , at this stage it is essential to isolate nempi as a physical effect under conditions that are as simple as possible .
we consider a domain of size @xmath2 in cartesian coordinates , @xmath3 , with periodic boundary conditions in the @xmath4 and @xmath5 directions and stress - free perfectly conducting boundaries at top and bottom ( @xmath6 ) .
the volume - averaged density is therefore constant in time and equal to its initial value , @xmath7 .
we solve the equations for the velocity @xmath8 , the magnetic vector potential @xmath9 , and the density @xmath10 , @xmath11 @xmath12 @xmath13 where @xmath14 is kinematic viscosity , @xmath15 is magnetic diffusivity , @xmath16 is the magnetic field , @xmath17 is the imposed uniform field , @xmath18 is the current density , @xmath19 is the vacuum permeability , @xmath20 is the traceless rate of strain tensor , and commas denote partial differentiation .
the forcing function @xmath21 consists of random , white - in - time , plane non - polarized waves with an average wavenumber @xmath22 , where @xmath23 is the lowest wavenumber in the domain .
the forcing strength is such that the turbulent rms velocity is approximately independent of @xmath24 with @xmath25 .
the gravitational acceleration @xmath26 is chosen such that @xmath27 , which leads to a density contrast between bottom and top of @xmath28 . here , @xmath29 is the density scale height .
our simulations are characterized by the fluid reynolds number @xmath30 , the magnetic prandtl number @xmath31 and the magnetic reynolds number @xmath32 . following earlier work @xcite
, we choose @xmath33 and @xmath34 in the range @xmath35@xmath36 .
the magnetic field is expressed in units of the local equipartition field strength near the top , @xmath37 , while @xmath38 is specified in units of the averaged value , @xmath39 .
we monitor @xmath40 , where @xmath41 is an average over @xmath5 and a certain time interval @xmath42 .
time is expressed in eddy turnover times , @xmath43 .
occasionally , we also consider the turbulent - diffusive timescale , @xmath44 , where @xmath45 is the estimated turbulent magnetic diffusivity .
another diagnostic quantity is the rms magnetic field in the @xmath46 fourier mode , @xmath47 , which is here taken as an average over @xmath48 , and is close to the top at @xmath49 .
( note that @xmath47 does not include the imposed field @xmath38 at @xmath50 . )
( averaged over @xmath48 ) on @xmath34 for @xmath51 . ]
the simulations are performed with the pencil code , which uses sixth - order explicit finite differences in space and a third - order accurate time stepping method .
we use numerical resolutions of @xmath52 and @xmath53 mesh points when @xmath54 , and @xmath55 when @xmath56 . to capture mean - field effects on the slower turbulent - diffusive timescale , which is @xmath57 times slower than the dynamical timescale , we perform simulations for several thousand turnover times .
the nempi phenomenon is already quite pronounced at intermediate values of @xmath58 ; see figure [ pbymxztn ] , where we show @xmath59 averaged over @xmath5 and @xmath60 ( denoted by @xmath41 ) at selected times during the first 2000 turnover times .
the @xmath34 dependence of @xmath61 is shown in figure [ psummaryr ] . for @xmath62 ,
the @xmath34 dependence is relatively weak , which suggests that nempi is a genuine high reynolds number effect .
the results for the case of @xmath63 show strong similarities to earlier mean - field simulations . during the first 500 turnover times , flux concentrations form first near the surface , but at later times the location of the peak magnetic field moves gradually downward .
this phenomenon is a direct consequence of the negative effective magnetic pressure , making such structures heavier than their surroundings .
their shape resembles that of a falling `` potato sack '' and has been seen in numerous mean - field calculations during the nonlinear stage of nempi @xcite . using the technique described in bkkr ,
we have found that for @xmath64 and @xmath65 , the effective magnetic pressure has a negative minimum . as expected from theory and mean - field calculations , nempi is only excited in a certain range of field strengths . in particular , only for @xmath66 between 0.02 and 0.2 do we see large - scale magnetic structures .
this is shown in figure [ pcomp_slice ] , where we see @xmath59 , again averaged over @xmath5 and a time interval @xmath67 , in which the field is statistically steady .
the clearest flux structure formations are seen for @xmath68 .
however , even for this case the flux concentrations are barely visible in a single snapshot .
this has been one of the reasons why nempi has not been noticed before in dns .
an additional handicap was that the simulations of bkkr used a smaller scale separation ratio of only 5 , which is nevertheless still sufficient for determining the governing mean - field coefficients and allows one to reach larger values of @xmath34 . vs. @xmath66 ( and vs. @xmath69 on the upper abscissa ) for @xmath70 .
] for @xmath51 , corresponding to @xmath71 , for @xmath70 ( solid line ) .
the growth rate is @xmath72 ( dotted line ) .
note that the total rms field , @xmath73 ( dotted line ) saturates much faster , as indicated by to upper abscissa . ] in figure [ psummary ] we plot @xmath61 as a function of @xmath66 , showing a peak at @xmath68 .
we recall that @xmath74 applies here to the location @xmath48 where @xmath47 has been evaluated , and there we have @xmath75 .
the fact that large - scale flux concentrations develop only for a certain range of imposed field strengths supports our interpretation that they are caused by nempi and not , for example , by some yet unknown dynamo mechanism . in all these cases
, @xmath76 grows rapidly and reaches a saturation field strength that is independent of @xmath77 , provided @xmath78 .
this suggests that this field is produced by small - scale dynamo action and not just by field line tangling .
another piece of evidence of the physical reality of nempi is shown in figure [ ppbymxztm ] , where we see that @xmath47 does indeed increase exponentially for the first 2000 turnover times , corresponding to about @xmath79 .
the growth rate is @xmath72 , which is much less than @xmath80 , but entirely compatible with mean - field calculations ( bkkr ; * ? ? ?
* ) . finally , to investigate the effects of the domain aspect ratio on the instability , we perform a calculation with @xmath51 , @xmath70 , and change @xmath81 to @xmath82 .
we find that the most unstable mode has a wavelength approximately equal to @xmath83 .
this result is also in agreement with mean - field models ( e.g. , fig .
14 of * ? ? ?
the large - scale flux concentrations have an amplitude of only @xmath84 and are therefore not seen in single snapshots , where the field reaches peak strengths comparable to @xmath74 .
furthermore , as for any linear instability , the flux concentrations form a repetitive pattern , but this might be an artifact of idealized conditions .
the present simulations have , for the first time , demonstrated conclusively that nempi can operate in hydromagnetic turbulence under proper conditions , namely , strong stratification , sufficient scale separation ( here @xmath85 ) , and a mean field in an optimal range ( here @xmath86 ; see figure [ psummary ] ) .
this instability has so far only been seen in mean - field simulations . by contrast , the present simulations are completely free of any mean - field consideration . the instability is argued to be a consequence of the reduction of turbulent hydromagnetic pressure by a mean magnetic field and can be understood as follows @xcite .
the combined reynolds and maxwell stress is @xmath87 , where @xmath88 and @xmath89 are velocity and magnetic fluctuations , respectively , and overbars indicate averaging . for isotropic turbulence , the turbulent hydromagnetic pressure is then @xmath90 . on the other hand , the total turbulent energy @xmath91 , is nearly conserved because a uniform mean magnetic field does not perform any work ; see @xcite for a numerical demonstration . the presence of an additional 1/2 factor in front of the @xmath92 term in the expression for @xmath93 , but not in that for @xmath94 , implies that the generation of magnetic fluctuations results in a reduction of @xmath93 , i.e. @xmath95 . for anisotropic turbulence
this negative contribution becomes larger .
this physical effect is independent of stratification , but to obtain an instability one needs strong stratification .
we speculate that , in the solar context , nempi plays a role in formation of active regions from mean fields generated by the solar dynamo .
let us now ask whether this instability alone can describe the formation of active regions at the solar surface . clearly , the flux concentrations we observe are not strong enough to be noticeable without averaging , while the active regions in the sun are seen without averaging .
this suggests that there may be additional mechanisms at work .
one possibility is that of the magnetic suppression of the convective heat flux that has been invoked to explain the formation of sunspots @xcite . when the mean magnetic field becomes larger than the equipartition field strength , i.e.
, when nempi does not work , and the characteristic spatial scale of the magnetic field is smaller than the density height , it is instead the parker magnetic buoyancy instability @xcite that is excited .
the presence of a vertical field might also have a strong effect .
indeed simulations of convection with an imposed vertical field have produced a segregation into magnetized and unmagnetized regions @xcite with formation of flux concentrations strong enough to be noticeable even without averaging .
on the other hand nempi might become more powerful at stronger stratification .
increased stratification clearly has an enhancing effect on the growth rate @xcite , but the effect on the saturation level has not yet been quantified .
furthermore , the interplay between nempi and the other effects also needs to be investigated .
our work has established a close link between what can be expected from mean - field studies and what actually happens in dns .
this correspondence is particularly important because dns can not reach solar parameters in any conceivable future .
hence a deeper understanding of solar convection can only emerge by studying mean - field models on the one hand and to determine turbulent mean - field coefficients from simulations on the other hand .
this concerns not only the dependence of the mean - field coefficients on parameters such as magnetic reynolds and prandtl numbers and scale separation ratio , but also the details of the source of turbulence .
in particular , it has already been shown that the negative effective magnetic pressure effect is not unique to forced turbulence , but it also occurs in turbulent convection @xcite , and thus in an unstably stratified layer .
we emphasize that the present work demonstrates the predictive power of mean - field theory at an advanced level where quasi - linear theory fails @xcite but the spectral @xmath1 approach @xcite has proven useful .
more work using mean - field models is needed to elucidate details of the mechanism of nempi .
for example , naive thinking suggests that the onset of nempi should occur at the depth where the effective magnetic pressure is minimum , but both mean - field models and dns show that this is not the case . at least at early times
, nempi appears most pronounced at the top of the domain , while the effective magnetic pressure is usually most negative at the bottom . on the other hand ,
the instability is a global one and local considerations such as these are not always meaningful .
another question is what happens when the imposed field is replaced by a dynamo - generated one . in that case
, the turbulence may be helical and new terms involving current density can occur in the expression for the mean - field stress . again , such possibilities are best studied using first the mean - field approach .
we acknowledge the nordita dynamo programs of 2009 and 2011 for providing a stimulating scientific atmosphere .
computing resources provided by the swedish national allocations committee at the center for parallel computers at the royal institute of technology in stockholm and the high performance computing center north in ume .
this work was supported in part by the european research council under the astrodyn research project no . | we present the first demonstration of the negative effective magnetic pressure instability in direct numerical simulations of stably stratified , externally forced , isothermal hydromagnetic turbulence in the regime of large plasma beta . by the action of this instability , an initially uniform horizontal magnetic field forms flux concentrations whose scale is large compared to the turbulent scale .
we further show that the magnetic energy of these large - scale structures is only weakly dependent on the magnetic reynolds number , provided its value is large enough for the instability to be excited .
our results support earlier mean - field calculations and analytic work which identified this instability .
applications to the formation of active regions in the sun are discussed . |
online social networks ( e.g. , facebook , google@xmath0 ) have become increasingly important resources for interacting with people , processing information and diffusing social influence . understanding and modeling the mechanisms by which these networks evolve are therefore fundamental issues and active areas of research .
the classical _ link prediction problem _ @xcite has attracted particular interest . in this
setting , we are given a snapshot of a social network at time @xmath1 and aim to predict links ( e.g. , friendships ) that will emerge in the network between @xmath1 and a later time @xmath2 . alternatively , we can imagine the setting in which some links existed at time @xmath1 but are missing at @xmath2 . in online social networks ,
a change in privacy settings often leads to missing links , e.g. , a user on google@xmath0might decide to hide her family circle between time @xmath1 and @xmath2 .
the missing link problem has important ramifications as missing links can alter estimates of network - level statistics @xcite , and the ability to infer these missing links raises serious privacy concerns for social networks .
since the same algorithms can be used to predict new links and missing links , we refer to these problems jointly as link prediction .
another problem of increasing interest revolves around node attributes @xcite .
many real - world networks contain rich categorical node attributes , e.g. , users in google@xmath0have profiles with attributes including employer , school , occupation and places lived . in the _ attribute inference problem _ , we aim to populate attribute information for network nodes with missing or incomplete attribute data .
this scenario often arises in practice when users in online social networks set their profiles to be publicly invisible or create an account without providing any attribute information .
the growing interest in this problem is highlighted by the privacy implications associated with attribute inference as well as the importance of attribute information for applications including people search and collaborative filtering . in this work ,
we simultaneously use network structure and node attribute information to improve performance of both the link prediction and the attribute inference problems , motivated by the observed interaction and homophily between network structure and node attributes .
the principle of social influence @xcite , which states that users who are linked are likely to adopt similar attributes , suggests that network structure should inform attribute inference
. other evidence of interaction @xcite shows that users with similar attributes , or in some cases antithetical attributes , are likely to link to one another , motivating the use of attribute information for link prediction .
additionally , previous studies @xcite have empirically demonstrated those effects on real - world social networks , providing further support for considering both network structure and node attribute information when predicting links or inferring attributes .
however , the algorithmic question of how to simultaneously incorporate these two sources of information remains largely unanswered .
the relational learning @xcite , matrix factorization and alignment @xcite based approaches have been proposed to leverage attribute information for link prediction , but they suffer from scalability issues .
more recently , backstrom and leskovec @xcite presented a supervised random walk ( srw ) algorithm for link prediction that combines network structure and edge attribute information , but this approach does not fully leverage node attribute information as it only incorporates node information for neighboring nodes .
for instance , srw can not take advantage of the common node attribute san francisco of @xmath3 and @xmath4 in fig .
[ figure : san ] since there is no edge between them .
yin et al .
@xcite proposed the use of _ social - attribute network _
( san ) to gracefully integrate network structure and node attributes in a scalable way .
they focused on generalizing random walk with restart ( rwwr ) algorithm to the san model to predict links as well as infer node attributes . in this paper , we generalize several leading supervised and unsupervised link prediction algorithms @xcite to the san model to both predict links and infer missing attributes .
we evaluate these algorithms on a novel , large - scale google@xmath0dataset , and demonstrate performance improvement for each of them .
moreover , we make the novel observation that inferring attributes could help predict links , i.e. , link prediction accuracy is further improved by first inferring missing node attributes .
in our problem setting , we use an undirected graph @xmath5 to represent a social network , where edges in @xmath6 represent interactions between the latexmath:[$n = have categorical attributes for nodes . for instance , in the google@xmath0social network , nodes are users , edges represent friendship ( or some other relationship ) between users , and node attributes are derived from user profile information and include fields such as employer , school , and hometown . in this work
we restrict our focus to categorical variables , though in principle other types of variables , e.g. , live chats , email messages , real - valued variables , etc . , could be clustered into categorical variables via vector quantization , or directly discretized to categorical variables .
we use a binary representation for each categorical attribute .
for example , various employers ( e.g. , google , intel and yahoo ) and various schools ( e.g. , berkeley , stanford and yale ) are each treated as separate binary attributes . hence , for a specific social network , the number of distinct attributes @xmath8 is finite ( though @xmath8 could be large ) .
attributes of a node @xmath9 are then represented as a @xmath8-dimensional trinary column vector @xmath10 with the @xmath11 entry equal to @xmath12 when @xmath9 has the @xmath11 attribute ( _ positive attribute _ ) , @xmath13 when @xmath9 does not have it ( _ negative attribute _ ) and @xmath14 when it is unknown whether or not @xmath9 has it ( _ missing attribute _ ) .
we denote by @xmath15 $ ] the attribute matrix for all nodes .
note that certain attributes ( e.g. female and male , age of 20 and 30 ) are mutually exclusive .
let @xmath16 be the set of all pairs of mutually exclusive attributes .
this set constrains the attribute matrix @xmath17 so that no column contains a @xmath12 for two mutually exclusive attributes .
we define the link prediction problem as follows : let @xmath18 and @xmath19 be snapshots of a social network at times @xmath20 and @xmath21 .
then the link prediction problem involves using @xmath22 to predict the social network structure @xmath23 . when @xmath24 , new links are predicted . when @xmath25 , missing links are predicted . in this paper
, we work with three snapshots of the google@xmath0network crawled at three successive times , denoted @xmath26 , @xmath27 and @xmath28 .
to predict new links , we use various algorithms to solve the link prediction problem with @xmath29 and @xmath30 and first learn any required hyperparameters by performing grid search on the link prediction problem with @xmath31 and @xmath32 . similarly , to predict missing links , we solve the link prediction problem with @xmath29 and @xmath33 and learn hyperparameters via grid search with @xmath34 and @xmath32 . for
any given snapshot , several entries of @xmath17 will be zero , corresponding to missing attributes . the attribute inference problem , which involves only a single snapshot of the network , is defined as follows : let @xmath35 be a snapshot of a social network .
then the attribute inference problem is to infer whether each zero entry of @xmath17 corresponds to a positive or negative attribute , subject to the constraints listed in @xmath16 .
our goal is to design scalable algorithms leveraging both network structure and rich node attributes to address these problems for real - world large - scale networks .
_ social - attribute network _
was first proposed by yin et al .
@xcite to predict links and infer attributes
. however , their original model did nt consider negative and mutually exclusive attributes . in this section ,
we review this model and extend it to incorporate negative and mutex attributes . given a social network @xmath36 with @xmath8 distinct categorical attributes , an attribute matrix @xmath17 and mutex attributes set @xmath16 , we create an augmented network by adding @xmath8 additional nodes to @xmath36 , with each additional node corresponding to an attribute . for each node @xmath9 in @xmath36 with positive or negative attribute @xmath37
, we create an undirected link between @xmath9 and @xmath37 in the augmented network . for each mutually exclusive attribute pair @xmath38
, we create an undirected link between @xmath37 and @xmath39 .
this augmented network is called the _ social - attribute network _ ( san ) since it includes the original social network interactions , relations between nodes and their attributes and mutex links between attributes . nodes in the san model corresponding to nodes in @xmath36 are called _ social nodes _ , and nodes representing attributes are called _
attribute nodes_. links between social nodes are called _ social links _ , and links between social nodes and attribute nodes are called _
attribute links_. attribute link @xmath40 is a _ positive attribute link _ if @xmath37 is a positive attribute of node @xmath9 , and it is a _ negative attribute link _ otherwise .
links between mutually exclusive attribute nodes are called _
mutex links_. intuitively , the san model explicitly describes the sharing of attributes across social nodes as well as the mutual exclusion between attributes , as illustrated in the sample san model of fig .
[ figure : san ] .
moreover , with the san model , the link prediction problem reduces to predicting social links and the attribute inference problem involves predicting attribute links .
we also place weights on the various nodes and edges in the san model .
these node and edge weights describe the relative importance of individual nodes or relationships across nodes and can also be used in a global fashion to balance the influence of social nodes versus attribute nodes and social links versus attribute links .
we use @xmath41 and @xmath42 to denote the weight of node @xmath9 and the weight of link @xmath43 , respectively . additionally , for a given social or attribute node @xmath9 in the san model , we denote by @xmath44 and @xmath45 respectively the set of _ all neighbors _ and the set of _ social neighbors _ connected to @xmath9 via social links or positive attribute links .
we define @xmath46 and @xmath47 in a similar fashion .
this terminology will prove useful when we describe our generalization of leading link prediction algorithms to the san model in the next section .
the fact that no social node can be linked to multiple mutex attributes is encoded in the _ mutex property _
, i.e. , there is no triangle consisting of a mutex link and two positive attribute links in any social - attribute network , which enforces a set of constraints for all attribute inference algorithms . in this work ,
we focus primarily on node attributes . however , we note that the san model can be naturally extended to incorporate _
edge attributes_. indeed , we can use a function ( e.g. , the logistic function ) to map a given set of attributes for each edge ( e.g. , edge age ) into the real - valued edge weights of the san model .
the attributes - to - weight mapping function can be learned using an approach similar to the one proposed by backstrom and leskovec @xcite .
link prediction algorithms typically compute a probabilistic score for each candidate link and subsequently rank these scores and choose the largest ones ( up to some threshold ) as putative new or missing links . in the following ,
we extend both unsupervised and supervised algorithms to the san model .
furthermore , we note that when predicting attribute links , the san model features a post - processing step whereby we change the lowest ranked putative positive links violating the mutex property to negative links .
liben - nowell and kleinberg @xcite provide a comprehensive survey of unsupervised link prediction algorithms for social networks .
these algorithms can be roughly divided into two categories : local - neighborhood - based algorithms and global - structure - based algorithms . in principle , all of the algorithms discussed in @xcite can be generalized for the san model . in this work we focus on representative algorithms from both categories and we describe below how to generalize them to the san model to predict both social links and attribute links .
we add the suffix ` -san ' to each algorithm name to indicate its generalization to the san model . in our presentation of the unsupervised algorithms ,
we only consider positive attribute links , though many of these algorithms can be extended to signed networks @xcite . + * common neighbor ( cn - san ) * is a local algorithm that computes a score for a candidate social or attribute link @xmath43 as the sum of weights of @xmath9 and @xmath48 s common neighbors , i.e. @xmath49 .
conventional cn only considers common social neighbors .
+ * adamic - adar ( aa - san ) * is also a local algorithm . for a candidate social link @xmath43
the aa - san score is @xmath50 conventional aa , initially proposed in @xcite to predict friendships on the web and subsequently adapted by @xcite to predict links in social networks , only considers common social neighbors .
aa - san weights the importance of a common neighbor proportional to the inverse of the log of social degree .
intuitively , we want to downweight the importance of neighbors that are either i ) social nodes that are social hubs or ii ) attribute nodes corresponding to attributes that are widespread across social nodes .
since in both cases this weight depends on the social degree of a neighbor , the aa - san weight is derived based on social degree , rather than total degree .
in contrast , for a candidate attribute link @xmath40 , the attribute degree of a common neighbor does influence the importance of the neighbor . for instance , consider two social nodes with the same social degree that are both common neighbors of nodes @xmath9 and @xmath37 . if the first of these social nodes has only two attribute neighbors while the second has @xmath51 attribute neighbors , the importance of the former social node should be greater with respect to the candidate attribute link .
thus , aa - san computes the score for candidate attribute link @xmath40 as @xmath52 * low - rank approximation ( lra - san ) * takes advantage of global structure , in contrast to cn - san and aa - san .
denote @xmath53 as the @xmath54 weighted social adjacency matrix where the @xmath55 entry of @xmath53 is @xmath42 if @xmath43 is a social link and zero otherwise .
similarly , let @xmath56 be the @xmath57 weighted attribute adjacency matrix where the @xmath58 entry of @xmath56 is @xmath59 if @xmath40 is a positive attribute link and zero otherwise .
we then obtain the weighted adjacency matrix @xmath60 for the san model by concatenating @xmath53 and @xmath56 , i.e. , @xmath61 $ ] .
the lra - san method assumes that a small number of latent factors ( approximately ) describe the social and attribute link strengths within @xmath60 and attempts to extract these factors via low - rank approximation of @xmath60 , denoted by @xmath62 .
the lra - san score for a candidate social or attribute link @xmath63 is then simply @xmath64 , or the @xmath65 entry of @xmath62 .
lra - san can be computed efficiently via truncated singular value decomposition ( svd ) . +
* cn + low - rank approximation ( cn+lra - san ) * is a mixture of local and global methods , as it first performs cn - san using a san model and then performs low - rank approximation on the resulting score matrix . after performing cn - san ,
let @xmath66 be the resulting @xmath54 score matrix for all social node pairs and @xmath67 be the resulting @xmath68 score matrix for all social - attribute node pairs .
by virtue of the cn - san algorithm , note that @xmath66 includes attribute information and @xmath67 includes social interactions .
- san then predicts social links by computing a low - rank approximation of @xmath66 denoted @xmath69 , and each entry of @xmath69 is the predicted social link score .
similarly , @xmath70 is a low - rank approximation of @xmath67 , and each entry of @xmath70 is the predicted score for the corresponding attribute link.$ ] , approximating @xmath71 with @xmath72 and using the @xmath65 entry of @xmath72 as a score for link @xmath63 . ] + * aa + low - rank approximation(aa+lra - san ) * is identical to cn+lra - san but with the score matrices @xmath66 and @xmath67 generated via the aa - san algorithm .
+ * random walk with restart ( rwwr - san ) * @xcite is a global algorithm . in the san model ,
a random walk with restart @xcite starting from @xmath9 recursively walks to one of its neighbors @xmath1 with probability proportional to the link weight @xmath73 and returns to @xmath9 with a fixed restart probability @xmath74 .
the probability @xmath75 is the stationary probability of node @xmath48 in a random walk with restart initiated at @xmath9 . in general , @xmath76 . for a candidate social link @xmath43
, we compute @xmath75 and @xmath77 and let @xmath78 .
note that rwwr for link prediction in previous work @xcite computes these stationary probabilities based only on the social network . for a candidate attribute link @xmath40
, rwwr - san only computes @xmath79 , and @xmath79 is taken as the score of @xmath40 .
+ we finally note that for predicting social links , if we set the weights of all attribute nodes and all attribute links to zero and we set the weights of all social nodes and social links to one , then all the algorithms described above reduce to their standard forms described in @xcite . is an @xmath68 matrix of zeros , so the truncated svd of @xmath60 is equivalent to that of @xmath53 except for @xmath8 zeros appended to the right singular vectors of @xmath53 . ] in other words , we recover the link prediction algorithms on pure social networks .
link prediction can be cast as a binary classification problem , in which we first construct features for links , and then use a classifier such as svms or logistic regression .
in contrast to unsupervised attribute inference , negative attribute links are needed in supervised attribute inference . +
* supervised link prediction ( slp - san ) * for each link in our training set , we can extract a set of topological features @xmath80 ( e.g. cn , aa , etc . )
computed from pure social networks and the similar features @xmath81 computed from the corresponding social - attribute networks .
we explored 4 feature combinations : i ) slp - i uses only topological features @xmath80 computed from social networks ; ii ) slp - ii uses topological features @xmath80 as well as an aggregate feature , i.e. , the number of common attributes of the two endpoints of a link ; iii ) slp - san - iii uses topological features @xmath81 ; and iv ) slp - san - vi uses topological features @xmath80 and @xmath81 .
slp - san - iii and slp - san - vi contain the substring ` san ' because they use features extracted from the san model .
slp - i and slp - ii are widely used in previous work @xcite . +
* supervised attribute inference ( sai - san ) * recall that attribute inference is transformed to attribute link prediction with the san model .
we can extract a set of topological features for each positive and negative attribute link .
moreover , the positive attribute links are taken as positive examples while the negative attribute links are taken as negative examples .
hence , we can train a binary classifier for attribute links and then apply it to infer the missing attribute links . in many real - world networks , most node attributes are missing .
[ figure : node - attri ] shows the fraction of users as a function of the number of node attributes in google@xmath0social network .
from this figure , we see that roughly 70% of users have no observed node attributes .
hence , we will also investigate an iterative variant of the san model .
we first infer the top attributes for users without any observed attributes .
we then update the san model to include these predicted attributes and perform link prediction on the updated san model .
this process can be performed for several iterations .
google launched its new social network service named google@xmath0 in early july 2011 .
we crawled three snapshots of the google@xmath0 social network and their users profiles on july 19 , august 6 and september 19 in 2011 .
they are denoted as jul , aug and sep , respectively .
we then pre - processed the data before conducting link prediction and attribute inference experiments . + * preprocessing social networks * in google@xmath0 , users divide their social connections into circles , such as a family circle and a friends circle .
if user @xmath9 is in @xmath48 s circle , then there is a directed edge @xmath82 in the graph , and thus the google@xmath0dataset is a directed social graph .
we converted this dataset into an undirected graph by only retaining edges @xmath43 if both directed edges @xmath43 and @xmath82 exist in the original graph .
we chose to adopt this filtering step for two reasons : ( 1 ) bidirectional edges represent mutual friendships and hence represent a stronger type of relationship that is more likely to be useful when inferring users attributes from their friends attributes ( 2 ) we reduce the influence of spammers who add people into their circles without those people adding them back .
spammers introduce fictitious directional edges into the social graph that adversely influence the performance of link prediction algorithms .
+ * collecting attribute vocabulary * google@xmath0profiles include short entries about users such as occupation , employment , education , places lived , and gender , etc .
we use employment and education to construct a vocabulary of attributes in this paper .
we treat each distinct employer or school entity as a distinct attribute .
google@xmath0 has predefined employer and school entities , although users can still fill in their own defined entities .
due to users changing privacy settings , some profiles in jul are not found in aug and sep , so we use jul to construct our attribute vocabulary .
specifically , from the profiles in jul , we list all attributes and compute frequency of appearance for each attribute . our attribute vocabulary is constructed by keeping attributes with frequency of at least 3 . + * constructing social - attribute networks * in order to demonstrate that the san model leverages node attributes well , we derived social - attribute networks in which each node has some positive attributes from the above google@xmath0social networks and attribute vocabulary .
specifically , for an attribute - frequency threshold @xmath83 , we chose the largest connected social network from jul such that each node has at least @xmath83 distinct positive attributes .
we also found the corresponding social networks consisting of these nodes in snapshots aug and sep .
social - attribute networks were then constructed with the chosen social networks and the attributes of the nodes . specifically , we chose @xmath84 to construct 6 social - attribute networks whose statistics are shown in table [ table : net_stat ] .
each social - attribute network is named by concatenating the snapshot name and the attribute - frequency threshold .
for example , ` jul4 ' is the social - attribute network constructed using jul and @xmath85 .
these names are indicated in the first column of the table . in the crawled raw networks ,
some social links in jul@xmath20 are missing in aug@xmath20 and sep@xmath20 , where @xmath86 .
these links are missing due to one of two events occurring between the jul and aug or sep snapshots : 1 ) users block other users , or 2 ) users set ( part of ) their circles to be publicly invisible after which point they can not be publicly crawled .
these missed links provide ground truth labels for our experiments of predicting missing links . however , these missing links can alter estimates of network - level statistics , and can have unexpected influences on link prediction algorithms @xcite .
moreover , it is likely in practice that companies like facebook and google keep records of these missing links , and so it is reasonable to add these links back to aug@xmath20 and sep@xmath20 for our link prediction experiments .
the third column in table [ table : net_stat ] is the number of all social links after filling the missing links into aug@xmath20 and sep@xmath20 . the second column _ # soci links _ is used for experiments of predicting missing links , and column _ # all soci links _
is used for the experiments of predicting new links . from these two columns ,
the number of new links or missing links can be easily computed .
for example , if we use aug2 as training data and sep2 as testing data for link prediction , the number of new links is @xmath87 , which is computed with entries in column _ # all soci links_. if we use aug2 as training data and jul2 as testing data in predicting missing links , the number of missing links is @xmath88 , which is computed with corresponding entries in column _ # soci links _ and _ # all soci links_. .*statistics of social - attribute networks .
* [ cols="^,^,^,^,^,^",options="header " , ] [ table : net_stat_a ] [ table : net_stat ]
[ table : link - pre ] in our experiments , the main metric used is auc , area under the receiver operating characteristic ( roc ) curve , which is widely used in the machine learning and social network communities @xcite .
auc is computed in the manner described in @xcite , in which both positive and negative examples are required . in principle
, we could use new links or missing links as positive examples and all non - existing links as negative examples .
however , large - scale social networks tend to be very sparse , e.g. , the average degree is @xmath89 in sep2 , and , as a result , the number of non - existing links can be enormous , e.g. , sep2 has around @xmath90 non - existing links .
hence , computing auc using all non - existing links in large - scale networks is typically computationally infeasible .
moreover , the majority of new links in typical online social networks close triangles @xcite , i.e. , are hop-2 links . for instance , we find that @xmath91 of the newly added links in google+ are hop-2 links .
we thus evaluate our large network experiments using hop-2 link data as in @xcite , i.e. , new or missing hop-2 links are treated as positive examples and non - existing hop-2 links are treated as negative examples . in a social - attribute network , there are two categories of hop-2 links : 1 ) those with two endpoints sharing at least one common social node , and 2 ) those with two endpoints sharing only common attribute nodes .
local algorithms applied to the original social network are unable to predict hop-2 links in the second category .
thus , we evaluate only with respect to hop-2 links in the first category , so as not to give unfair advantage to algorithms running on the social - attribute network . to better understand whether the auc performance computed on hop-2 links can be generalized to performance on any - hop links
, we additionally compute auc using any - hop links on the smaller google@xmath0networks .
in general , different nodes and links can have different weights in social - attribute networks , representing their relative importance in the network . in all of our experiments in this paper
, we set all weights to be one and leave it for future work to learn weights .
we compare our link prediction algorithms with supervised random walk ( srw ) @xcite , which leverages edge attributes , by transforming node attributes to edge attributes .
specifically , we compute the number of common attributes of the two endpoints of each existing link . as in @xcite
, we also use the number of common neighbors as an edge attribute .
we adopt the wilcoxon - mann - whitney ( wmw ) loss function and logistic edge strength function in our implementations as recommended in @xcite .
we compare our attribute inference algorithms with two algorithms , baseline and link , introduced by zheleva and getoor @xcite . using only node attributes
, baseline first computes a marginal attribute distribution and then uses an attribute s probability as its score .
link trains a classifier for each attribute by flattening nodes as the rows of the adjacency matrix of the social networks .
zheleva and getoor @xcite found that link is the best algorithm when group memberships are not available .
we use svm as our classifier in all supervised algorithms . for link prediction
, we extract six topological features ( cn - san , aa - san , lra - san , cn+lra - san , aa+lra - san and rwwr - san ) from both pure social networks and social - attribute networks .
hence , slp - i , slp - ii , slp - san - iii and slp - san - vi use 6 , 7 , 6 and 12 features , respectively . for attribute inference
, we extract 9 topological features for each attribute link .
we adopt two ranks ( detailed in [ sec : attri - infer ] ) for each low - rank approximation based algorithms , thus obtaining 6 features .
the other three features are cn - san , aa - san and rwwr - san . to account for the highly imbalanced class distribution of examples for supervised link prediction and
attribute inference we downsample negative examples so that we have equal number of positive and negative examples ( techniques proposed in @xcite could be used to further improve the performance ) .
we use the pattern _
dataset1_-_dataset2 _ to denote a train - test or train - validation pair , with _
dataset1 _ a training dataset and _ dataset2 _ a testing or validation dataset .
when conducting experiments to predict new links on the aug@xmath20-sep@xmath20 train - test pair , srw , classifiers and hyperparameters of global algorithms , i.e. , ranks in lra - san , cn+lra - san , and aa+lra - san and the restart probability @xmath74 in rwwr - san , are learned on the jul@xmath20-aug@xmath20 train - validation pair .
similarly , when predicting missing links on train - test pair aug@xmath20-jul@xmath20 , they are learned on train - validation pair sep@xmath20-aug@xmath20 , where @xmath86 .
the cn - san and aa - san algorithms are implemented in python 2.7 while the rwwr - san algorithm and supervised random walk ( srw ) are implemented in matlab , and all of them are run on a desktop with a 3.06 ghz intel core i3 and 4 gb of main memory .
lra - san , cn+lra - san and aa+lra - san algorithms are implemented in matlab and run on an x86 - 64 architecture using a single 2.60 ghz core and 30 gb of main memory . in this section
we present evaluations of the algorithms on the google+ dataset .
we first show that incorporating attributes via the san model improves the performance of both unsupervised and supervised link prediction algorithms .
then we demonstrate that inferring attributes via link prediction algorithms within the san model achieves state - of - the - art performance .
finally , we show that by combining attribute inference and link prediction in an iterative fashion , we achieve even greater accuracy on the link prediction task . to demonstrate the benefits of combining node attributes and network structure , we run the san - based link prediction algorithms described in section [ sec : algorithms ] both on the original social networks and on the corresponding social - attribute networks ( recall that the san - based unsupervised algorithms reduce to standard unsupervised link prediction algorithms when working solely with the original social networks ) .
+ * predicting new links * table [ table : link - pre ] shows the auc results of predicting new links for each of our datasets .
we are able to draw a number of conclusions from these results .
first , the san model improves every unsupervised learning algorithm on every dataset , save for lra - san on aug2-sep2 .
second , table [ table : link - pre - svm ] shows that attributes also improve supervised link prediction performance since slp - san - vi , slp - san - iii and slp - ii outperform slp - i .
moreover , slp - san - vi , which adopts features extracted from both social networks and social - attribute networks , achieves the best performance , thus demonstrating the power of the san model .
third , comparing rwwr - san in table [ table : link - pre-4-any - hop ] and srw in table [ table : link - pre - svm ] , we observe that the san model is better than srw at leveraging node attributes since rwwr - san with attributes outperforms srw .
this result is not surprising given that srw is designed for edge attributes and when transforming node attributes to edge attributes , we lose some information .
for instance , as illustrated in fig .
[ figure : san ] , nodes @xmath3 and @xmath4 share the attribute san francisco . when transforming node attributes to edge attributes , this common attribute information is lost since
@xmath3 and @xmath4 are not linked .
[ figure : exp1-link - pre - roc-4 ] shows the roc curves of the cn+lra - san algorithm .
we see that curve of cn+lra - san with attributes dominates that of cn+lra - san without attributes , demonstrating the power of the san model to effectively incorporate the additional predictive information of attributes .
+ [ table : infer - link ] * predicting missing links * missing links can be divided into two categories : 1 ) links whose two endpoints have some social links in the training dataset .
2 ) links with at least one endpoint that has no social links in the training dataset .
category 1 corresponds to the scenarios where users block users or users set a part of their friend lists ( e.g. family circles ) to be private .
category 2 corresponds to the scenario in which users hide their entire friend lists .
note that all hop-2 missing links belong to category 1 .
in addition to performing experiments to show that the san model improves missing link prediction , we also perform experiments to explore which category of missing links is easier to predict .
table [ table : infer - link ] shows the results of predicting missing links on various datasets . as in the new - link prediction setting , the performance of every algorithm is improved by the san model , except for lra - san on aug4-jul4 and rwwr - san on aug4-jul4 for hop-2 missing links .
when comparing tables [ table : infer - missing - link-4-any - hop ] and [ table : infer - missing - link-4-any - hop-1 ] or tables [ table : infer - missing - link - svm ] and [ table : infer - missing - link - svm-1 ] , we conclude that the missing links in category 2 are harder to predict than those in category 1 .
rwwr - san without attributes performs poorly when predicting any - hop missing links in both categories ( as indicated by the entry with @xmath92 in table [ table : infer - missing - link-4-any - hop ] ) .
this poor performance is due to the fact that rwwr - san without attributes assigns zero scores for all the missing links in category 2 ( positive examples ) and positive scores for most non - existing links ( negative examples ) , making many negative examples rank higher than positive examples and resulting in a very low auc . in this section ,
we focus on inferring attributes using the san model . in our next set of experiments in section [ sec : iterate ] , we use the results of these attribute inference algorithms to further improve link prediction , and the results of this iterative approach further validate the performance of the san model for attribute inference . since the first step of iterative approach of section [ sec : iterate ] involves inferring the top attributes for each node , we employ an additional performance metric called pre@@xmath93 in our attribute inference experiments . compared to auc
, pre@@xmath93 better captures the quality of the top attribute predictions for each user .
specifically , for each sampled user , the top-@xmath93 predicted attributes are selected , and ( unnormalized ) pre@@xmath93 is then defined as the number of positive attributes selected divided by the number of sampled users .
we address score ties in the manner described in @xcite .
since most google@xmath0 users have a small number of attributes , we set @xmath94 in our experiments . when evaluating algorithms for the inference of missing attributes , we require ground truth data .
in general , ground truth for node attributes is difficult to obtain since it is often not possible to distinguish between negative and missing attributes .
however , for most users the number of attributes is quite small , and so we assume that users with many positive attributes have no missing attributes . hence , we evaluate attribute inference on users that have at least 4 specified attributes , i.e. , we work with users in sep4 and assume that each attribute link in sep4 is either positive or negative . in our experiment , we sample 10% of the users in sep4 uniformly at random , remove their attribute links from sep4 , and evaluate the accuracy with which we can infer these users attributes .
all removed positive attribute links are viewed as positive examples , while all the negative attribute links of the sampled users are treated as negative examples .
we run a variety of algorithms for attribute inference , and for each algorithm we average the results over 10 random trials . as noted above
, we evaluate the performance of attribute inference using both auc and pre@@xmath93 . for the low - rank approximation based algorithms ,
i.e. , lra - san , cn+lra - san and aa+lra - san , we report results using two different ranks , 100 and 1000 , and indicate which was used by the number following the algorithm name in fig .
[ figure : infer - attri ] .
we choose these two small ranks for computational reasons and also based on the fact that low - rank approximation methods assume that a small number of latent factors ( approximately ) describe the social - attribute networks . for rwwr - san
, we set the restart probability @xmath74 to be 0.7 .
[ figure : infer - attri ] shows the attribute inference results for various algorithms .
several interesting observations can be made from this figure .
first , under both metrics , all san - based algorithms perform better than baseline , save lra100-san and lra1000-san under pre@2,3,4 metric , which indicates that the san model is good at leveraging network structure to infer missing attributes .
second , we find that auc and pre@@xmath93 provide inconsistent conclusions about relative algorithm performance .
for instance , the mean auc values suggest that sai - san beats all other algorithms .
however , several unsupervised algorithms outperform sai - san with respect to pre@2,3,4 .
the inconsistencies between the two metrics are expected since auc is a global measurement while pre@@xmath93 is a local one .
our sai - san algorithm dominates link under both auc and pre@2,3,4 metrics , thus demonstrating the power of mapping attribute inference to link prediction with the san model .
section [ sec : link - prediction ] demonstrated that knowledge of a user s attributes can lead to significant improvements in link prediction .
however , in real - world social networks like google+ , the vast majority of user attributes are missing ( see fig .
[ figure : node - attri ] ) . to increase the realized benefits of social - attribute networks with few attributes , we propose first inferring missing attributes for each user whose
attributes are missing and then performing link prediction on the inferred social - attribute networks .
recall that sai - san achieves the best auc , rwwr - san achieves the best pre@@xmath93 in inferring attributes ( see fig . [
figure : infer - attri ] ) and aa - san achieves comparable pre@@xmath93 results while being more scalable . thus , in the following experiments , we use aa - san to first infer the top-@xmath93 missing attributes for users , and subsequently perform link prediction using various methods . in our experiments ,
when we are working on the pair _ train - test _ , we sample 10% of the users of _ train _ uniformly at random and remove their attributes .
we then run three variants of link prediction algorithms : i ) without attributes , ii ) with only the remaining attributes , and iii ) with the remaining attributes along with the inferred attributes .
the top-4 attributes are inferred for each sampled user by aa - san .
we report the results averaged over 10 trials .
the hyperparameters of the global algorithms are the same as those in ( section [ sec : link - prediction ] ) , which are learned from the corresponding train - validation pair .
table [ table : iter - link - pre ] shows the results of first inferring attributes and then predicting new links on the aug4-sep4 train - test pair .
table [ table : iter - infer - missing - link ] shows the results of first inferring attributes and then predicting missing links on the aug4-jul4 train - test pair .
we see that the inferred attributes improve the performance of all algorithms except lra - san on predicting missing links , which is unable to make use of attributes as demonstrated earlier in table [ table : infer - missing - link-4 ] .
the aucs obtained with inferred attributes for all other algorithms are very close to those obtained with all positive attributes as shown in table [ table : link - pre-4 ] .
this further demonstrates that aa - san is an effective algorithm for attribute inference .
a wide range of link prediction methods have been developed .
liben - nowell and kleinberg @xcite surveyed a set of unsupervised link prediction algorithms .
li @xcite proposed maximal entropy random walk ( merw ) .
lichtenwalter et al .
@xcite proposed the propflow algorithm which is similar to rwwr but more localized .
however , none of these approaches leverage node attribute information .
link prediction methods leveraging attribute information first appear in the relational learning community @xcite .
however , these approaches suffer from scalability issues .
for instance , the largest network tested in @xcite has about @xmath95 nodes . recently ,
backstrom and leskovec @xcite proposed the supervised random walk ( srw ) algorithm to leverage edge attributes .
however , srw does not handle the scenario in which two nodes share common attributes ( e.g. nodes @xmath3 and @xmath4 in fig .
[ figure : san ] ) , but no edge already exists between them .
mapping link prediction to a classification problem @xcite is another way to incorporate attributes .
we have shown that classifiers using features extracted from the san model perform very well .
yang et al .
@xcite proposed to jointly predict links and propagate node interests ( e.g. , music interest ) .
their algorithm relies on the assumption that each node interest has a set of explicit attributes . as a result
, their algorithm can not be applied to our scenario in which it s hard ( if possible ) to extract explicit attributes for our node attributes .
previous works in @xcite aim at inferring node attributes ( e.g. , ethnicity and political orientation ) using supervised learning methods with features extracted from user names and user - generated texts .
zheleva and getoor @xcite map attribute inference to a relational classification problem .
they find that methods using group information achieve good results .
these approaches are complementary to ours since they use additional information apart from network structure and node attributes . in this paper
, we transform the attribute inference problem into a link prediction problem with the san model .
therefore , any link prediction algorithm can be used to infer missing attributes .
more importantly , we demonstrate that attribute inference can in turn help link prediction with the san model .
we comprehensively evaluate the _ social - attribute network _ ( san ) model proposed in @xcite in terms of link prediction and attribute inference . more specifically
, we adapt several representative unsupervised and supervised link prediction algorithms to the san model to both predict links and infer attributes .
our evaluation with a large - scale novel google@xmath0network dataset demonstrates performance improvement for each of these generalized algorithm on both link prediction and attribute inference .
moreover , we demonstrate a further improvement of link prediction accuracy by using the san model in an iterative fashion , first to infer missing attributes and subsequently to predict links .
interesting avenues for future research include devising an iterative algorithm that alternates between attribute and link prediction , learning node and edge weights in the san model , and incorporating edge attributes , negative node attributes and mutex edges into large - scale experiments .
we would like to thank di wang , satish rao , mario frank , kurt thomas , and shobha venkataraman for insightful feedback .
this work is supported by the nsf under grants no .
ccf-0424422 , 0311808 , 0832943 , 0448452 , 0842694 , 0627511 , 0842695 , 0808617 , 1122732 , 0831501
ct - l , by the afosr under muri award no .
fa9550 - 09 - 1 - 0539 , by the afrl under grant no .
p010071555 , by the office of naval research under muri grant no .
n000140911081 , by the muri program under afosr grant no .
fa9550 - 08 - 1 - 0352 , the nsf graduate research fellowship under grant no .
dge-0946797 , the dod through the ndseg program , by intel through the istc for secure computing , and by a grant from the amazon web services in education program .
any opinions , findings , and conclusions or recommendations expressed in this material are those of the author(s ) and do not necessarily reflect the views of the funding agencies . | the effects of social influence and homophily suggest that both network structure and node attribute information should inform the tasks of link prediction and node attribute inference . recently ,
yin et al .
@xcite proposed _ social - attribute network _ ( san ) , an attribute - augmented social network , to integrate network structure and node attributes to perform both link prediction and attribute inference .
they focused on generalizing the random walk with restart algorithm to the san framework and showed improved performance . in this paper , we extend the san framework with several leading supervised and unsupervised link prediction algorithms and demonstrate performance improvement for each algorithm on both link prediction and attribute inference .
moreover , we make the novel observation that attribute inference can help inform link prediction , i.e. , link prediction accuracy is further improved by first inferring missing attributes .
we comprehensively evaluate these algorithms and compare them with other existing algorithms using a novel , large - scale google@xmath0dataset , which we make publicly available . |
the efforts of many observing teams over several decades have provided rich data concerning magnetic activity in stars of various spectral classes . however in many cases the general form of the activity of stars of a given spectral class remains debatable . in particular , observations find two distinct magnetic topologies for m - dwarfs : 1 ) strong ( kilogauss ) almost axisymmetric poloidal magnetic fields , and 2 ) considerably weaker nonaxisymmetric fields , sometimes including a substantial toroidal component ( morin et al .
stars with different topologies can have roughly the same rotation rates , mass , and other parameters .
this finding has been interpreted as dynamo bistability , i.e. as two different regimes of dynamo action being stable for the same set of stellar parameters ( morin et al .
2011a , b ; gastine et al .
however , two stable regimes of a dynamo in the same parameter range seems not to be very probable ( cf .
e.g. rdler et al . 1990 ; moss & sokoloff 2009 ) . on the other hand ,
chabrier & kker ( 2006 ) suggest that fully convective stars host a non - axisymmetric global dynamo ( such a possibility was also suggested in a slightly different context by , e.g. , barker & moss 1994 ) .
this dynamo is steady ( equatorial dipole drifting in longitude ) and therefore showing no reversals .
this model relies on the observational fact ( barnes et al . 2005 ) supported by numerical simulations ( kker & rdiger 2005 ) that differential rotation in fully convective stars is very small . as a result chabrier & kker ( 2006 ) neglect differential rotation and considered an @xmath0 dynamo .
the efficiency of differential rotation in winding up magnetic fields can be estimated by the dimensionless dynamo number @xmath1 where @xmath2 is the angular velocity variation within the convection zone , @xmath3 is the convection zone thickness and @xmath4 is the eddy magnetic diffusivity .
differential rotation modelling suggests that the decrease in @xmath2 with decreasing temperature can be compensated by a simultaneous decrease in @xmath4 , so that @xmath5 actually _ increases _ in cooler stars ( kitchatinov & olemskoy 2011 ; kitchatinov 2013 ) .
this means that the m - dwarfs have small but very efficient differential rotation and it is quite probable that they host oscillatory , axisymmetric mean - field dynamos .
taking all this into account , it seems reasonable to consider another possible explanation of the m - dwarf activity phenomenon .
we suggest that magnetic cycles in the form of oscillatory axisymmetric fields are present within the population of m - dwarfs , and that the stars that are observed to have weak nonaxisymmetric fields are observed at epochs of reversal , with the strong axisymmetric dipoles being present at the analogues of solar maxima .
there are strong observational indications for a difference in magnetic topologies between fully convective stars and stars with convective envelopes , with magnetic fields of fully convective stars being dominated by axisymmetric poloidal configurations ( gregory et al .
2012 ) . nevertheless , an analogy with the dynamics of the ( relatively weak ) solar poloidal field is possible and can be useful .
note that the solar activity is determined mainly by the toroidal magnetic field .
there is a time lag between the cyclic oscillations of toroidal and poloidal magnetic fields so that the solar magnetic dipole is strong during the minima of solar activity .
a straightforward verification of this explanation could be performed by monitoring of a sample of m - dwarfs over times exceeding the expected period .
estimates in the next section suggest however that magnetic cycles in m - dwarfs can last for several decades .
if m - dwarfs do have cycles , the cycles could therefore be substantially longer than that of the sun , while at the moment only a 3-year monitoring ( 2006 - 2009 ) is available ( morin et al .
2010 ) .
a straightforward objection to the last explanation could be that solar mean - field dynamo models driven by differential rotation and mirror - asymmetric turbulence ( @xmath6-dynamos ) give axisymmetric mean magnetic fields , and the magnetic dipole has to vanish during its inversion , and be parallel to the rotation axis between its reversals .
recent progress in understanding solar observations and the solar dynamo ( moss et al .
2013a , b ) have provided a new understanding of the reversals of the solar magnetic dipole .
this allows us to elaborate the idea under discussion quantitatively , and to suggest a way to verify it that does not ultimately require a long - term monitoring programme ( which of course still remains highly desirable ) .
taking into account the inherent uncertainties and arbitrariness of dynamo modelling for wide classes of stars , and that it is far from clear how generic any particular model ( set - up , choice of parameters , etc ) can be , nevertheless we performed an exploratory modelling to establish the possibility of oscillatory dynamos existing in m - dwarfs , as follows .
we computed the differential rotation for a star with @xmath7 , rotating with period of 10 days , using the numerical mean - field model of kitchatinov & olemskoy ( 2011 ) .
the result is shown in fig . [ diff ] .
such a star is fully convective , but the model requires an ( in this case artificial ) inner boundary of the convection zone to be imposed at @xmath8 .
the model does not prescribe eddy transport coefficients but estimates them from the entropy gradient , so that the @xmath5 parameter of eq .
( [ c_omega ] ) can also be estimated .
taking the turbulent prandtl number , @xmath9 to have the value unity gives @xmath10 , which is not small in spite of the small differential rotation in fig .
the eddy viscosity @xmath11@xmath12s@xmath13 used in this estimate is taken at the middle radius @xmath14 ( @xmath15 mm ) .
thus the diffusive time is @xmath16 years .
if the cycle time is close to the diffusion time , as it is for the sun , very long cycles can be expected .
the circulation time for the meridional flow is also long , @xmath1760 yr ( the typical flow velocity is 10 cm / s except in the thin surface boundary layer , where it is about 4 m / s ) . thus the cycle time for an advection - dominated dynamo will also be long .
we use these estimates to obtain the reference values for the dynamo governing parameters for m - dwarfs and experiment with numbers around these reference quantities .
[ dwarfdyn ] demonstrates the results for the simplest dynamo model with uniform diffusion , @xmath18 uniform with radius and @xmath19 dependence on co - latitude , isotropic diffusion and alpha .
the model estimates the threshold value of the dimensionless parameter , @xmath20 for onset of dynamo together with the field structure and oscillation frequency .
the eddy viscosity is estimated by the differential rotation code .
we take the magnetic prandtl number of eq .
( [ p_m ] ) as a free parameter .
we conclude from this limited modelling that an oscillatory dynamo model with solar type behaviour can be considered as a viable option at least at the present level of investigation .
we stress that further modelling of dynamo action in m - dwarfs remains highly desirable . the transition from steady to oscillatory dynamos with increasing @xmath21 , seen in fig .
[ dwarfdyn ] , indicates that the dynamo operates in the @xmath22 regime . in this regime
, poloidal fields can be expected to be much stronger than in solar - type @xmath23 dynamos .
equipartition poloidal fields of kilogauss strength @xmath24 g can be expected for the fully convective m - dwarfs ( @xmath25 g@xmath26 , @xmath27 cms@xmath13 at @xmath14 from the model of the @xmath28 star structure ) .
we start by noting that many solar observers ( e.g. antonucci 1974 ; zhukov & veselovsky 2000 ; livshits & obridko 2006 ) have reported that the solar magnetic dipole does not vanish during the reversal .
recently de rosa et al .
( 2012 ) presented a comprehensive data sample for the two last solar activity cycles which convincingly demonstrate that this is indeed the case .
moss et al .
( 2013a ) suggested how to resolve this apparent contradiction between expectations from dynamo modelling and observation .
the point is that a mean - field dynamo model deals with _
mean _ magnetic field and the averaging is performed over an ensemble of convective velocity cells , while the observational magnetic dipole data refer to _ large - scale _ magnetic field .
both quantities coincide for an infinitely large ensemble of convective cells , but in practice the number of cells is only moderately large ( @xmath29 is a crude relevant estimate , ( see moss et al .
2013a for details ) .
because the convective cell ensemble contains a not extremely large number of cells , large - scale fluctuations of magnetic field arise which yield a fluctuating component @xmath30 of the solar magnetic dipole @xmath31 , of order @xmath32 here @xmath33 is the r.m.s .
value of small - scale magnetic field , i.e. the magnetic fluctuations , @xmath34 is the typical value of the mean magnetic field which is determined mainly by the toroidal magnetic field @xmath35 , and the factor @xmath36 takes into account that the magnetic dipole moment is determined by the poloidal magnetic field @xmath37 .
moss et al . ( 2013a ) demonstrated that the estimate for the solar magnetic dipole at the epoch of inversion , based on eq .
( [ estim ] ) and the available information about the relevant solar parameters is consistent with this picture .
moss et al .
( 2013b ) compared this scenario with observational data more systematically , to learn that the interval during which the fluctuating part of the magnetic dipole is larger than the part determined by the mean magnetic field is about 4 months , i.e. about 3% of the solar magnetic cycle .
[ migr ] shows ( from an example from the moss et al .
2013b analysis ) the migration of the dipole from one solar pole to the other , and clearly demonstrates the random nature of the reversal .
assuming that the magnetic activity of m - dwarfs is more - or - less similar to that of the sun , we can use eq .
( [ estim ] ) to determine the relative time @xmath38 during the magnetic activity cycle during which the magnetic dipole is determined by magnetic fluctuations and is strongly inclined to the rotation axis .
because the scenario assumes that at that time the axisymmetric dipolar field is minimal we conclude that observations taken at that epoch would identify this star as exhibiting the second type of activity .
if we consider a random sample of m - dwarfs then the value @xmath38 immediately gives the relative number @xmath39 of m - dwarfs in the sample which demonstrates the second type of activity .
the temporal evolution of the axisymmetric solar magnetic dipole is approximately sinusoidal ( moss et al .
2013a ) , i.e. much simpler than the evolution of the mean sunspot number . assuming that the same is valid for m - dwarfs , and taking into account eq .
( [ estim ] ) we can say what physical parameters are required , e.g. @xmath39 , to get a satisfactory correspondence with the observational data . in particular ,
if we want to explain that about 30% of m - dwarfs demonstrate the second type of activity , we have to assume that the number of convective cells at the surface of m - dwarfs is about two order of magnitudes lower than near the surface of the sun , i.e. @xmath40 instead of @xmath41 .
( given the much greater relative depth of the convection zone in m - dwarfs compared to the sun , an increase in the size of convection cells is not implausible . ) then , eq .
( [ estim ] ) reads that @xmath42 increases by factor @xmath43 in comparison with the sun and because the oscillations of the dipole strength are near to sinusoidal , the time at which fluctuations are weaker than the mean value of the dipole grows by a factor @xmath40 , and now becomes about 1/3 of the cycle duration . as a result , the relative number of m - dwarfs which demonstrate the second type of activity increases by a factor @xmath43 , giving 30% .
we have suggested a scenario which explains why observations identify two types of magnetic activity in m - dwarfs .
we propose that a m - dwarf can either be observed at a time far from that of magnetic field inversion ( giving the first type of activity ) , or near the inversion ( giving the second type ) .
a quantification of the scenario under discussion can be performed on the basis of eq .
( [ estim ] , ) by determination of the percentage of m - dwarfs that demonstrate the second type of activity .
we stress that such quantification does not require an effort - consuming long - term monitoring of m - dwarf activity .
of course , such a monitoring remains highly desirable for further understanding of stellar magnetic activity .
even a crude estimate of the relative number of m - dwarfs with the second type of activity will provide in our framework a perspective for monitoring activity in m - dwarfs .
if , say , this number were about 10% , then it would only be necessary to perform monitoring during 1/10 of a stellar cycle in order to observe a transition from one activity type to the other one in a given object .
another problem of interest would be to determine how this relative number varies between solar type stars and m - dwarfs .
indeed , any star with periodic dynamo driven large - scale fields might be expected to display similar behaviour , but whether this could be detected by current observational techniques is another matter .
ds and llk are grateful to the russian foundation for basic research for financial support under grants 12 - 02 - 00170 and 13 - 02 - 00277 respectively .
morin , j. , delfosse , x. , donati , j .- f . ,
dormy , e. , forveille , t. , jardine , m. m. , petit , p. , schrinner , m. 2011a , in sf2a-2011 : proc .
meeting french soc .
astrophys . , g. alecian , k. belkacem , r. samadi & d. valls - gabaud ( eds ) , 503 . | m - dwarfs demonstrate two types of activity : 1 ) strong ( kilogauss ) almost axisymmetric poloidal magnetic fields ; and 2 ) considerably weaker nonaxisymmetric fields , sometimes including a substantial toroidal component .
dynamo bistability has been proposed as an explanation .
however it is not straightforward to obtain such a bistability in dynamo models . on the other hand ,
the solar magnetic dipole at times of magnetic field inversion becomes transverse to the rotation axis , while the magnetic field becomes weaker at times far from that of inversion .
thus the sun resembles a star with the second type of activity .
we suggest that m - dwarfs can have magnetic cycles , and that m - dwarfs with the second type of activity can just be stars observed at times of magnetic field inversion
. then the relative number of m - dwarfs with the second type of activity can be used in the framework of this model to determine parameters of stellar convection near the surface .
sun : activity stars : activity stars : magnetic fields magnetic fields dynamo stars : late - type [ firstpage ] |
the free monoid @xmath0 of words over the alphabet @xmath1 ordered by the prefix relation has a @xmath2-regular rooted tree structure in which the empty word is the root and the words of length @xmath3 constitute the level @xmath3 in the tree .
the @xmath2 children of the vertex @xmath4 are the vertices @xmath5 , for @xmath6 .
denote this @xmath2-regular rooted tree by @xmath7 .
any automorphism @xmath8 of @xmath7 can be ( uniquely ) decomposed as @xmath9 , where @xmath10 is called the _ root permutation _ of @xmath8 , and @xmath11 , @xmath6 , are tree automorphisms , called the ( first level ) _ sections _ of @xmath8 .
the root permutation @xmath12 and the sections @xmath13 are determined uniquely by the relation @xmath14 , for all @xmath15 and @xmath16 . the action of a tree automorphism can be extended to an isometric action on the boundary @xmath17 consisting of the infinite words over @xmath18 .
the space @xmath19 is a compact ultrametric space homeomorphic to a cantor set . for any permutation @xmath20 in @xmath21 define a @xmath2-ary tree automorphism @xmath22 by @xmath23 , where @xmath24 is the identity automorphism if @xmath25 is in the support of @xmath20 and @xmath26 otherwise .
the action of the automorphism @xmath27 on @xmath7 is given ( recursively ) by @xmath28 _ hanoi towers group on @xmath2 pegs _ , @xmath29 , is the group @xmath30 of @xmath2-ary tree automorphisms generated by the automorphisms @xmath27 , @xmath31 , corresponding to the transpositions in @xmath21 .
the group @xmath32 derives its name from the fact that it models the well known hanoi towers problem on @xmath2 pegs ( see @xcite ) . in this problem ,
@xmath3 disks of distinct sizes , enumerated @xmath33 by their size , are placed on @xmath2 pegs , denoted @xmath34 .
any configuration of disks is allowed as long as no disk is placed on top of a smaller disk . a legal move consists of moving the top disk from one peg to the top of another peg ( as long as the new configuration is allowed ) .
the @xmath3-disk configurations are in bijective correspondence with the @xmath35 words of length @xmath3 over @xmath1 .
namely , the word @xmath36 over @xmath18 corresponds to the unique configuration in which the disk @xmath25 , @xmath37 , is placed on peg @xmath38 .
the action of the automorphism @xmath27 corresponds to a move between the pegs @xmath25 and @xmath39 .
the group @xmath32 , @xmath29 , is an example of an automaton group . in general ,
an _ invertible automaton _ is a quadruple @xmath40 in which @xmath41 is a finite set of states , @xmath18 a finite alphabet , @xmath42 a _ transition function _ and @xmath43 an _ output function _ such that , for each state @xmath44 , the restriction @xmath45 is a permutation in @xmath46 ( see @xcite ) .
the states of @xmath47 define recursively tree automorphisms by setting the permutation @xmath48 to be the root permutation of @xmath49 and the state @xmath50 to be the section @xmath51 of @xmath49 at @xmath52 .
the group of tree automorphisms @xmath53 generated by the automorphisms corresponding to the states of the invertible automaton @xmath47 is called the _ automaton group _ of @xmath47 .
invertible automata are often represented by diagrams such as the one on the left in figure [ figure ] corresponding to @xmath54 .
each state @xmath49 is labeled by the permutation @xmath48 and the labeled edges describe the transition function ( if @xmath55 then there is an edge labeled @xmath52 connecting @xmath49 to @xmath56 ) .
an automaton group is _ contracting _ if the length ( with respect to the generating set @xmath41 ) of each section @xmath13 , @xmath57 , of @xmath8 is shorter than the length of @xmath8 , for all sufficiently long elements @xmath8 .
a spherically transitive ( transitive on all levels ) group of tree automorphisms @xmath58 is _ regularly branching _ over its subgroup @xmath59 if @xmath59 is a normal subgroup of finite index in @xmath58 such that @xmath60 ( @xmath2 factors ) is a normal subgroup of finite index in @xmath58 that is geometrically contained in @xmath59 ( meaning that the @xmath2 factors act independently on the corresponding @xmath2 subtrees of @xmath7 rooted at the vertices of the first level of @xmath7 ; see @xcite ) .
the hanoi towers group @xmath61 is contracting and regularly branching over its commutator .
the groups @xmath32 act spherically transitively on @xmath7 , but are not contracting for @xmath62 .
let @xmath47 be an automaton , @xmath63 the corresponding automaton group , @xmath64 an infinite geodesic ray starting at the root of @xmath7 , @xmath65 the prefix of @xmath64 of length @xmath3 , @xmath66 the stabilizer @xmath67 and @xmath68 the stabilizer @xmath69 , @xmath70 .
denote by @xmath71 ( or just @xmath72 ) the schreier graph @xmath73 and by @xmath74 the schreier graph @xmath75 , where @xmath41 is the generating set defined by the states of @xmath40 .
we assume that the action of @xmath58 on @xmath7 is spherically transitive .
thus the graphs @xmath74 are connected , have size @xmath35 , @xmath70 , and @xmath58 is infinite .
the graph @xmath74 is indeed the graph of the action of @xmath58 on level @xmath3 in the tree and @xmath72 is the graph of the action on the orbit of @xmath64 in @xmath19 . the sequence of graphs @xmath76 converges to the infinite graph @xmath72 in the space of pointed graphs @xcite ( based at @xmath65 , @xmath70 , and @xmath64 , respectively ) .
for example , in the case of the hanoi group on 3 pegs @xmath61 the schreier graph @xmath77 is given in figure [ figure ] ( the automorphisms @xmath78 , @xmath79 and @xmath80 are denoted by @xmath81 , @xmath82 and @xmath83 , respectively in the figure ) .
the group @xmath61 has been independently constructed in @xcite as a group whose limit space is the sierpiski gasket .
all our asymptotic considerations are intended in the following sense . for two functions ,
we write @xmath84 if there exists @xmath85 such that @xmath86 , for all @xmath87 , and @xmath88 if @xmath84 and @xmath89 .
denote by @xmath90 the growth function of @xmath72 counting the number of vertices in @xmath72 that are in the ball of radius @xmath3 centered at @xmath64 .
the growth of @xmath72 is either exponential ( @xmath91{\gamma(n ) } = c>1 $ ] ) , intermediate ( @xmath91{\gamma(n ) } = 1 $ ] , @xmath92 , for all @xmath93 ) or polynomial ( @xmath91{\gamma(n ) } = 1 $ ] , @xmath94 , for some @xmath93 ) .
the growth is polynomial in any schreier graph @xmath72 associated to a contracting group @xcite .
for example , @xmath95 in case of @xmath61 , and the growth is linear for the first group from @xcite and the erschler group @xcite .
the growth of @xmath72 is exponential for the automata generating the lamplighter group @xcite and baumslag - solitar solvable groups @xcite , the bellaterra automaton @xcite generating the free product of three cyclic groups of order 2 , and the aleshin automaton @xcite generating the free group of rank 3 .
thus the schreier graph @xmath72 is amenable , for all hanoi towers groups ( another example providing schreier graphs of intermediate growth is the group studied in @xcite ) .
it is not known if , for @xmath62 , @xmath32 is amenable or if it contains free subgroups of rank 2 ( @xmath61 is amenable but not elementary amenable ) .
denote by @xmath99 the diameter of the graph @xmath74 .
the growth of the diameter function @xmath99 is either exponential , intermediate or polynomial .
since we assumed spherical transitivity of the action , we have that the schreier graphs @xmath74 are connected and @xmath100 .
this implies that if either @xmath99 or @xmath90 grows polynomially then the other one grows exponentially .
thus , contracting groups provide examples with exponential diameter growth . as a concrete example
, it is well known that @xmath101 for @xmath61 .
it can be shown that examples of polynomial ( even linear ) diameter growth are given by the the realizations of the lamplighter group in @xcite and by baumslag - solitar solvable groups @xmath102 , @xmath103 , in @xcite .
thus the hanoi towers groups provide examples with intermediate diameter growth ( for @xmath62 ) .
the classical hanoi towers problem asks for the number of steps needed to reach the configuration @xmath105 from @xmath106 .
it can be shown that this distance is asymptotically equal ( it is not equal ! ) to the diameter @xmath99 .
the frame - stewart algorithm ( see @xcite ) solves the hanoi towers problem in @xmath107 steps and it follows from the work of szegedy @xcite that this is asymptotically optimal .
let @xmath108 be the adjacency matrix of @xmath74 and @xmath109 be the corresponding markov operator ( related to the simple random walk on @xmath74 ) .
similarly , let @xmath110 be the adjacency matrix of @xmath72 and @xmath111 the corresponding markov operator .
the _ spectrum _ of @xmath74 ( or @xmath72 ) is , by definition , the spectrum of @xmath112 ( or @xmath111 ) .
the regularity of @xmath74 implies that 1 is in the spectrum of @xmath74 .
let @xmath113 be the _ spectral gap _
@xmath114 , where @xmath115 is the largest eigenvalue of @xmath74 different from 1 .
the family @xmath76 is , by definition , a family of _ expanders _ if there is a constant @xmath116 such that @xmath117 , for all @xmath3 .
if @xmath76 do not form a family of expanders then the spectral gap tends to 0 at a rate that is exponential ( @xmath118{\delta(n)}=c<1 $ ] ) , intermediate ( @xmath119{\delta(n)}= 1 $ ] , @xmath120 , for all @xmath93 ) or polynomial ( @xmath118{\delta(n)}= 1 $ ] , @xmath121 , for some @xmath93 ) .
chung inequality tells us that , for a graph @xmath122 , the diameter @xmath123 , where @xmath124 is the largest absolute value of an eigenvalue of @xmath72 different from 1 . in our situation @xmath125 and by using spectrum shifting ( achieved by adding loops ) we obtain @xmath126 . in case of an expander family @xmath127 , for some constant @xmath85
, i.e. the diameter growth is linear .
the converse is not true ( the lamplighter and the baumslag - solitar examples have linear diameter growth ) .
examples of automaton groups producing either expanders or intermediate decay of the spectral gap are not known yet .
examples with exponential decay are provided by the first group in @xcite and @xmath61 .
more generally , for contracting groups the spectral gap decays exponentially since the diameters grow exponentially .
an example with polynomial decay is provided by the lamplighter group @xcite .
the rate of convergence of @xmath128 to 0 is of special interest and is related to the behavior of the kns spectral measure of @xmath72 and growth of flner sets .
the kns measure @xmath129 is limit of counting measures @xmath130 on @xmath74 @xcite ( @xmath131 is the ratio @xmath132 , where @xmath133 counts the eigenvalues of @xmath74 in @xmath134 ) .
we present here a full description of the spectrum of each 3-regular schreier graph @xmath74 modeling the hanoi towers problem on 3 pegs as well as the spectrum of the 3-regular limit graph @xmath96 and the associated kns spectral measure .
let @xmath58 be the hanoi towers group on three pegs @xmath61 .
the spectrum of @xmath74 , as a set , has @xmath135 elements and is equal ( after re - scaling by 3 ) to @xmath136 where @xmath137 is the polynomial @xmath138 .
the multiplicity of the @xmath139 eigenvalues in @xmath140 , @xmath141 is @xmath142 and the multiplicity of the @xmath143 eigenvalues in @xmath144 , @xmath145 is @xmath146 , where @xmath147 and @xmath148 , for @xmath149 .
the spectrum of @xmath96 , as a set , is equal ( after re - scaling by 3 ) to @xmath150 it consists of a set of isolated points @xmath151 and its set of accumulation points @xmath152 , which is the julia set of the polynomial @xmath137 and is a cantor set .
the kns spectral measure is discrete , concentrated on @xmath153 and the measure of the eigenvalues in @xmath154 is @xmath155 , @xmath156 .
the method used to calculate the above spectra is a continuation of the direction set in @xcite . at the core of this method
lies a renormalization principle .
the action of @xmath61 on level @xmath3 of @xmath7 induces @xmath157-dimensional permutation matrix representation @xmath158 .
the spectra of @xmath74 , i.e. of the adjacency matrices @xmath159 , are then calculated by considering two - dimensional pencils @xmath160 of matrices ( where @xmath161 is the identity matrix and @xmath162 is additional matrix aiding the computation ) .
the zeros of @xmath163 are related for various @xmath3 by using iterations of a two - dimensional rational map @xmath164 .
the spectrum of @xmath96 is the intersection of an invariant set of @xmath165 ( consisting of a family of hyperbolae ) with the line @xmath166 .
the rational map @xmath165 is semi - conjugate to the one - dimensional polynomial map @xmath167 above , which makes possible the complete calculation of the spectra .
a similar result is obtained ( by using a different method that also has the renormalization spirit in its root ) in @xcite for the spectrum of the infinite @xmath168-regular ( with single exception ) graph approximating the sierpiski gasket ( the spectrum of this graph , as a set , was known since the 1980 s from the work of bllissard ) .
the spectrum is obtained by iterations of a quadratic polynomial that is not conjugate to @xmath137 . on the other hand , iterations of a polynomial conjugate to @xmath137 appear in @xcite in the functional equation for the green s function of the graph obtained by gluing two copies of the graph from @xcite along a vertex .
laurent bartholdi and zoran unik . some solvable automaton groups . in _
topological and asymptotic aspects of group theory _ ,
volume 394 of _ contemp .
_ , pages 1130 .
soc . , providence , ri , 2006 . r. grigorchuk and a. uk . on the asymptotic spectrum of random walks on infinite families of graphs .
in _ random walks and discrete potential theory _ , sympos .
math . , xxxix , pages 188204 .
cambridge univ . press , cambridge , 1999 .
mario szegedy . in how many steps
the @xmath2 peg version of the towers of hanoi game can be solved ? in _ stacs 99 ( trier ) _ , volume 1563 of _ lecture notes in comput .
_ , pages 356361 .
springer , berlin , 1999 . | we present relations between growth , growth of diameters and the rate of vanishing of the spectral gap in schreier graphs of automaton groups .
in particular , we introduce a series of examples , called hanoi towers groups since they model the well known hanoi towers problem , that illustrate some of the possible types of behavior . |
early generations of stars affect the subsequent evolution of the universe in diverse ways , such as the cosmic reionization and metal - enrichment of the intergalactic medium .
in contrast to a number of theoretical attempts to understand their nature in the last decade ( e.g. , bromm & larson 2004 ; ciardi & ferrara 2005 ) , their observational confirmations are thus far extremely limited . among them , surveys of metal - poor stars in the galactic halo provide unique tools to catch a glimpse of star formation in very low - metallicity environments in early galaxies ( e.g. , beers & christlieb 2005 ) . in recent years
, hundreds of extremely metal - poor ( emp ) stars ( [ fe / h]@xmath6 ) have been found in the halo .
a high fraction of them is noticeably enhanced with surface carbon abundance .
such carbon - enhanced extremely metal - poor ( cemp ) stars have attracted particular attention in constraining the stellar initial mass function ( imf ) in metal - poor environments . especially , those with @xmath7-process element enhancement , cemp-@xmath7 stars , are considered to originate from secondary stars in binary systems .
they acquired the surface carbon during the agb phase of intermediate - mass ( several @xmath8 ) primary stars through the binary - mass transfer .
although such carbon enhanced stars are present among population i stars as well ( about 1 % ) , the cemp-@xmath7 stars make up far higher fraction ( about 20 % ) of the emp stars ( e.g. , lucatello et al . 2005 ) .
since such a high fraction can not be accounted for under the same imf as in the solar vicinity , lucatello et al .
( 2005 ) concluded that the imf is biased toward higher mass in the early galaxy . by using more sophisticated stellar evolution models , komiya et al .
( 2007 ) reached similar conclusion .
although there is apparent scarcity of stars in the ultra - metal poor ( ump ) range ( @xmath9[fe / h]@xmath10 ) thus far ( e.g. , salvadori et al .
2007 ) , three stars has been discovered with [ fe / h]@xmath11 and all of them are carbon - enhanced ( christlieb et al . 2002 ; frebel et al . 2005 ; norris et al .
suda et al .
( 2004 ) showed that their carbon enhancement can be explained by the binary - transfer hypothesis similar to that for the cemp stars . extending
the above argument to lower metallicity range , tumlinson ( 2007 ) concluded that the imf is even more skewed toward massive ones .
it is quite natural to conceive that this trend persists also in the ump range .
moreover , the current lacking of ump stars might suggest that even less low - mass stars are formed in this range .
is there any possible mechanism that favors the formation of intermediate - mass stars in these metallicity ranges ? in metal - free gas ,
the first stars are expected to be very massive , more than @xmath12 , owing to inefficient fragmentation of dense cores whose mass scale is @xmath13 formed at @xmath14 ( bromm , coppi , & larson 1999 ) . on the other hand , a slight amount of dust
enables fragmentation at higher densities and then low - mass ( @xmath15 ) core formation owing to the cooling by dust thermal emission ( schneider et al .
2002 , 2003 , 2006 ; omukai et al . 2005 ; tsuribe & omukai 2006 ( paper i ) ) . in this case , although the upper limit of stellar mass is not known , there is no particular reason favoring the formation of intermediate - mass stars . here
, we propose a physical mechanism that prohibits low - mass star formation in the metallicity range @xmath16}_{0 } \la -4 $ ] despite being above the critical metallicity @xmath17}_{\rm cr}$ ] for dust - induced fragmentation.}_{0 } \equiv { \rm log}(z / z_{\sun})$ ] .
] this allows transition from intermediate - mass to low - mass star formation mode around the ump to emp range provided that a significant fraction of metals are already locked into dust by the time metallicity reaches this value .
fragmentation properties of star - forming cores are largely determined by their thermal evolution . in figure 1
, we show the temperature per unit hydrogen nuclei @xmath18 @xmath19 for metallicities @xmath17_{0}}=-5.5 , -4.5 $ ] , and -3.5 ( solid lines ) .
also those with @xmath20_{0 } } \le -3.8 $ ] are shown by dotted lines with an increment of 0.1 .
these are calculated using a one - zone model as in omukai et al .
( 2005 ) , except for the dust model , for which we use that produced in the pair - instability supernova of a metal - free @xmath21 star ( schneider et al .
dust from type ii sn gives a similar result ( see schneider et al .
2006 ) . in the case of metallicity
[ m / h]@xmath22 -4.5 , a sudden jump in @xmath18 is present around @xmath23 , which is caused by heat injection due to three - body h@xmath2 formation ( palla , salpeter , & stahler 1983 ) . this temperature jump is observed only in the range @xmath24_{0 } } \la -4.0 $ ] ( see dotted lines ) . for lower metallicities ( e.g. , [ m / h]@xmath25 ) , the temperature is already high at the density where the three - body reaction begins and the consequence of heat injection is just a gradual increase in temperature . on the other hand , for higher metallicities ( e.g. , [ m / h]@xmath26 ) ,
h@xmath2 formation is almost complete , owing to dust surface reactions .
thus , the three - body reaction plays minor role and the heat injection is small . only around [ m / h]@xmath27 does heat injection by three - body reaction result in a dramatic temperature increase .
we expect that when the equation of state becomes stiff owing to the h@xmath2 formation heating , a hydrostatic core forms and that it becomes spherical to a high degree .
since an initial elongation of the core of @xmath28 is required for fragmentation in the dust - cooling phase ( paper i ) , this heating is expected to prevent the subsequent fragmentation . in the following section ,
we demonstrate that this is indeed the case by way of hydrodynamical calculations .
we study the evolution of cloud cores at high densities where the effects of the heating by three - body h@xmath2 formation and cooling by dust are important by performing hydrodynamical calculations .
we use a barotropic equation of state ( eos ) in figure 1 for thermal evolution . as in paper
i , we choose the self - similar solution with @xmath29 as an unperturbed initial state , to which both non - spherical density / velocity and random - velocity perturbations are added in the same manner as in paper i. the initial amplitude of elongation @xmath30 is taken to be unity , where @xmath31 is the short ( long , respectively ) core axis length . in paper
i , we started calculation just before the onset of the dust cooling without rotation . here to see the effect of h@xmath2 formation heating , we start calculation at lower initial number density of @xmath32 at the center . as well as non - rotating cores , in this paper the cores with initial rigid rotation are considered .
total mass of the cloud is @xmath33 , where @xmath34 depends on metallicity ( figure 1 ) .
we used the smoothed particle hydrodynamics ( sph ) based on the godunov - type scheme ( tsuribe & inutsuka 1999 ; inutsuka 2002 ) as in paper i , but with implementation of particle splitting by the condition that smoothing length is to be smaller than @xmath35 of the jeans length . in particle
splitting , eight daughter particles are created on the vertices of a cube ( martel et al .
2006 ) in high density regions .
the initial number of sph particles is @xmath36 .
we remark that our method meets the resolution condition for physical fragmentation ( e.g. , klein et al .
2004 ) .
the evolution of the core with metallicity @xmath17_{0}}=-4.5 $ ] is shown in figure 2 ( a1-a4 ) .
the initial core has angular velocity of @xmath37 , where @xmath38 is the initial density at the center . as a result of the prominent h@xmath2 formation heating at @xmath39 , the pressure force exceeds gravity .
the collapse is almost halted along the short axis of the core while it continues along the long axis .
consequently , the elongation of the core diminishes as can be seen in figure 3 . at this time
, the core becomes almost hydrostatic and the accretion shock appears on the surface ( figure 2 ( a2 ) ) .
the core continues to contract owing to accretion of the ambient medium .
although the core becomes elongated during the dust - cooling phase ( @xmath40 ) , for @xmath41 the eos becomes adiabatic and the growth of elongation slows down ( see figure 3 ) . during the dust - cooling phase , elongation grows at the disk formation epoch ( @xmath42 ) .
however , with a small value of elongation at the onset of dust cooling , the maximum elongation reaches at most @xmath43 ( figure 2 ( a3 ) ; figure 3 ) .
we calculated the cases with other values of initial angular velocity ( @xmath44 , and @xmath45 ) and found that the maximum value of elongation does not exceed @xmath46 in any case .
since this falls below the critical value for fragmentation @xmath47 ( paper i ) , fragmentation does not occur .
we followed the evolution of the core until the early accretion stage and confirmed that it does not fragment ( see figure 2 ( a4 ) ) .
therefore , the minimum fragmentation mass is set by the jeans mass before the h@xmath2 formation heating , which is @xmath48 . here
we show that both the cores with @xmath17_{0}}=-5.5 $ ] and -3.5 fragment in the dust - cooling phase .
we illustrate the case of @xmath17_{0}}=-5.5 $ ] with initial angular velocity @xmath37 in figure 2 ( b1-b4 ) .
for @xmath40 the temperature drops rapidly due to the dust cooling ( figure 1 ) , and the core becomes elongated to a filament ( figure 2 ( b2,b3 ) ; figure 3 ) .
although the temperature evolution becomes adiabatic after the core becomes optically thick , for @xmath49 , the elongation continues to increase owing to inertia . at @xmath50 ,
the elongation attains its maximum of @xmath51 , which is enough for the filament to fragment ( paper i ) . as seen in figure 2 ( b4 )
, the filament fragments into five pieces and an additional five density peaks are still developing . in the non - rotating case ,
the maximum elongation is @xmath52 ( figure 3 ) , and two fragments are observed along with additional five growing density peaks .
that is , rotation enhances the elongation and promote the fragmentation of the cores .
this is because the elongation has longer time to grow during the collapse delayed by rotation .
we also calculated the case of higher metallicity @xmath17_{0}}=-3.5 $ ] , and found similar evolutionary features , although with larger maximum elongation of @xmath53 ( figure 3 ) even without rotation . the core fragmented into at least seven pieces .
the dust cooling produces low - mass fragments of star - forming cores with metallicity as low as [ m / h]@xmath54 , consistent with previous studies .
however , there is a pocket in the metallicity range @xmath16_{0 } } \la -4 $ ] , where the low - mass fragmentation is prohibited by sudden heat injection due to the three - body h@xmath2 formation . in this range ,
the fragmentation mass is @xmath55 .
no low - mass ( @xmath56 ) core is formed , although fragmentation of this intermediate - mass core into a binary is not rejected .
once the metallicity exceeds @xmath17_{0 } } \sim -4 $ ] , low - mass star formation becomes possible by the dust cooling without hindrance by the three - body h@xmath2 formation heating .
this provides a physical mechanism for the high characteristic mass ( @xmath57 several @xmath8 ; tumlison 2007 ) of low - metallicity stars , which is suspected from the observed high frequency of low - metallicity carbon - enhanced stars . in addition , the observed scarcity of ultra metal - poor stars , if confirmed , can be explained .
this also means that we need to modify the simplest scenario for the transition in star formation modes in the metal - poor gas ; i.e. , below a critical metallicity @xmath17_{\rm cr } } ( \simeq -6 ... -5)$ ] only massive stars are formed while low - mass stars are always formed in gas with metallicities above it ( e.g. , salvadori et al .
2006 ) . in this paper
, we showed that h@xmath2 formation on dust grains quenches the sudden heat injection by three - body h@xmath2 formation at a metallicity [ m / h]@xmath58 and enables low - mass star formation above this metallicity .
the range of metallicity @xmath24_{0 } } \la -4.0 $ ] , where the sudden temperature jump appears , is slightly wider , but similar to the width of the observed gap in the mdf .
although this coincidence is encouraging , it should not be taken at the face value since our calculation still contains uncertain parameters .
in particular , the nature of dust , e.g. , gas - dust ratio , composition , and size distribution , in the early universe is very uncertain . in the early universe
, the dust is assumed to be produced in supernova events and indeed a signature of dust has been observed in the extinction of high - z quasars and grbs ( maiolino et al . 2004 ; stratta et al .
2007 ) . on the other hand ,
in the nearby universe , there is no firm evidence for the dust production in supernova remnants ( e.g. , bouchet et al .
even if the dust production is efficient , a significant fraction of it might be destroyed by a reverse shock on the spot ( bianchi & schneider 2007 ; nozawa et al .
in addition , the rate coefficient of three - body reaction rate is still uncertain , in particular , at low temperatures ( flower & harris 2007 ; glover 2007 ) . considering those uncertainties , we should admit that although our prediction that the heating by three - body reaction prevents low - mass fragmentation in some metallicity range somewhere around [ m / h]@xmath27 remains valid qualitatively , the exact metallicity range where this mechanism works would be altered . finally , we discuss effects of rotation .
our result shows that the rotation accelerates the growth of elongation , thereby promoting the fragmentation .
this is consistent with the result by clark , glover , & klessen ( 2007 ) , who showed that , even in the primordial gas , where no fragmentation is observed without rotation , rotating cores can fragment after the formation of rotation supported disks although fragmentation is far less frequent than other cases . in our cases ,
the core does not fragment in the metallicity range @xmath16}_{0 } \la -4 $ ] even with rotation .
since parameters of the initial cores are not thoroughly surveyed here , some cores might result in fragmentation in this metallicity range .
even so , we expect that the number of such low - mass fragments is smaller than in other range .
more studies on the effects of rotation on low - metallicity star formation are awaited .
we thank simon glover , raffaella schneider and an anonymous referee for comments improving the manuscript .
numerical computations were in part carried out on vpp5000 systems at the cfca of naoj and the altix 3700 at the yukawa insutitute ( kyoto ) .
this research was supported in part by grants - in - aid from mext of japan ( ko;18740117,18026008 ) .
beers , t. c. , & christlieb , n. 2005 , , 43 , 531 bianchi , s. , & schneider , r. 2007 , , 378 , 973 bouchet , p. , dwek , e. , danziger , j. , arendt , r. g. , de buizer , i. j. m. , park , s. , suntzeff , n. b. , kirshner , r. p. , & challis , p. 2006
, , 650 , 212 bromm , v. & larson , r. b. 2004 , , 42 , 79 bromm , v. & loeb , a. 2003 , , 425 , 812 christlieb , n. , bessell , m. s. , beers , t. c. , gustafsson , b. , korn , a. , barklem , p. s. , karlsson , t. , mizuno - wiedner , m. , & rossi , s. 2002 , , 419 , 904 ciardi , b. , & ferrara , a. 2005 , space science reviews , 116 , 625 clark , p. c. , glover , s. c. o. , & klessen , r. s. 2007 , , submitted , ( preprint arxiv0706.0613 [ astro - ph ] ) frebel , a. et al .
2005 , , 434 , 871 flower , d. r. & harris , g. j. 2007 , , 377 , 705 glover , s. c. o. 2007 , first star iii conference proceedings , arxiv:0708.3086 [ astro - ph ] inutsuka , s. 2002 , jcoph , 179 , 238 jappsen , a .- k . ,
glover , s. c. o. , klessen , r. s. , & mac low , m .- m .
2007 , first star iii conference proceedings , arxiv:0708.4363 [ astro - ph ] klein , r. i. , fisher , r. & mckee , c. f. 2004 , rev .
conf . , 22 , 3 komiya , y. , et al .
2007 , , 658 , 367 lucatello , s. , gratton , r. g. , beers , t. c. , & carretta , e. 2005 , , 625 , 833 martel , h. , evans , n. j. , ii , & shapiro , p. r. 2006 , , 163 , 122 maiolino , r. , schneider , r. , oliva , e. , bianchi , s. , ferrara , a. , mannucci , f. , pedani , m , & roca sogorb , m. 2004 , , 431 , 533 norris , j. e. 2007 , , submitted ( preprint arxiv:0707.2657 [ astro - ph ] ) omukai , k. , tsuribe , t. , schneider , r. & ferrara , a. 2005 , , 626 , 627 palla , f. , salpeter , e. e. , & stahler , s. w. 1983 , , 271 , 632 salvadori , s. , schneider , r. & ferrara , a. 2006 , , submitted ( preprint astro - ph/0611130 ) santoro , f. , & shull , j. m. 2006 , , 643 , 26 schneider , r. , ferrara , a. , natarajan , p. , & omukai , k. 2002 , , 571 , 30 schneider , r. , ferrara , a. , salvaterra , r. , omukai , k , & bromm , v. 2003 , , 422 , 869 schneider , r. , inoue , a. k. , omukai , k. & ferrara , a. 2006 , , 369 , 1437 stratta , g. , maiolino , r. , fiore , f. , & delia , v. 2007 , , 661 , l9 suda , t. , aikawa , m. , machida , m. n. , fujimoto , m. y. , & iben , icko , jr . 2004 , , 611 , 476 tratta , g. , maiolino , r. , fiore , f. , & delia , v. 2007 , , 661 , l9 tsuribe , t. & inutsuka , s. 1999 , , 523 , l155 tsuribe , t. & omukai , k. 2006 , , 642 , l61 ( paper i ) tumlinson , j. 2007 , , 665 , 1361 | if a significant fraction of metals is in dust , star - forming cores with metallicity higher than a critical value @xmath0 are able to fragment by dust cooling , thereby producing low - mass cores . despite being above the critical metallicity , a metallicity range is found to exist around @xmath1 where low - mass fragmentation is prohibited . in this range ,
three - body h@xmath2 formation starts at low ( @xmath3 100k ) temperature and thus the resulting heating causes a dramatic temperature jump , which makes the central part of the star - forming core transiently hydrostatic and thus highly spherical . with little elongation , the core does not experience fragmentation in the subsequent dust - cooling phase .
the minimum fragmentation mass is set by the jeans mass just before the h@xmath2 formation heating , and its value can be as high as @xmath4 . for metallicity higher than @xmath5 , h@xmath2 formation
is almost completed by the dust - surface reaction before the onset of the three - body reaction , and low - mass star formation becomes possible .
this mechanism might explain the higher characteristic mass of metal - poor stars than in the solar neighborhood presumed from the statistics of carbon - enhanced stars . |
in a series of papers gefen , mandelbrot and aharony @xcite examined the possibility of phase transitions for discrete - symmetry spin models ( e.g. the ising model ) on self - similar fractal structures . by applying renormalization - group techniques they found that the critical properties of such systems vary with several topological characteristics : the ( noninteger ) fractal dimensionality @xmath0 , the order of ramification @xmath1 and lacunarity @xcite .
they established that a lower critical dimensionality is not defined and found that the ising model on systems with given @xmath0 have transition temperature @xmath20 if the minimum order of ramification @xmath3 is finite , and @xmath40 if @xmath3 is infinite .
the order of ramification @xmath1 at a point @xmath5 of a structure is defined as the number of bonds which one must cut in order to isolate an arbitrarily large bounded set of points connected to @xmath5 . for two - dimensional sierpinski gasket ( sg ) @xmath3=3 and @xmath6=4 @xcite .
subsequently , monceau and hsiao @xcite did a monte carlo study of the ising model on fractals with infinite @xmath3 which had the same fractal dimension 1@xmath73 but different structure .
they found direct evidence of so - called weak universality in such systems , i.e. the critical exponents depend not only on @xmath0 , the symmetry of the order parameter and the range of interactions but also on topological features of the fractal , in particular on lacunarity which measures the deviation from translational symmetry .
the question remained as to what happens in the case of @xmath8-component spin models , with @xmath92 , on fractal structures .
in @xcite a correspondence between pure resistor network connecting sites of a given lattice and the low - temperature properties of the spins with continuous symmetry ( @xmath92 ) on the same lattice @xcite was used to conclude that there is no long - range order at any finite temperature if the fractal dimension @xmath102 even in the case of infinite order of ramification . for two - dimensional sg @xmath113/@xmath122=1.585 .
subsequently , in their study of the effect of phase fluctuations of the superconducting order parameter in two - dimensional sierpinski gasket wire networks @xcite vallat , korshunov and beck @xcite examined the xy - model on such a lattice in the harmonic approximation .
they found that the energy of a vortex is always finite on the two - dimensional sierpinski gasket implying that there is no berezinskii - kosterlitz - thouless transition @xcite associated with vortex - antivortex unbinding as free vortices are always present .
here we present a monte carlo study of the xy - model on two - dimensional sierpinski gaskets described by the hamiltonian @xmath13 where 0@xmath142@xmath15 is the angle variable on site @xmath16 , @xmath17 denotes the nearest neighbors and @xmath180 is the coupling constant . in the case of xy - model on periodic and quasiperiodic lattices
@xcite the most important excitations at low temperatures are the spin - waves and the harmonic approximation to ( [ ham ] ) used in @xcite is quite good for the purpose of studying the low - temperature behavior .
we compute the heat capacity , the helicity modulus and susceptibility for sierpinski gaskets of several sizes .
the size dependence of the helicity modulus and its sensitivity to the boundary conditions clearly indicate that in the thermodynamic limit there is no phase transition at finite temperature .
the rest of the paper is organized as follows . in section [ numerics ]
we describe numerical procedure used in calculations and present our numerical results . in section [ summary ]
we summarize our conclusions .
the procedure which we used to generate two - dimensional sierpinski gaskets is illustrated in figure 1 , which shows the transition from the zeroth - order sierpinski gasket ( the equilateral triangle of side @xmath19 ) to the first order gasket .
our basic unit was the fourth order sierpinski gasket with 123 sites ( a gasket of order @xmath20 has @xmath21=3(3@xmath22 + 1)/2 sites ) and the list of the nearest neighbors of each site .
the higher order sierpinski gaskets were created as subsequent generations of the fourth order gasket and we considered gaskets up to order @xmath20=7 with 3282 sites . in going from the gasket of order @xmath20 to the gasket of order @xmath20 + 1 the list of the nearest neighbors for sites close to the three corners of each of the three @xmath20th order gaskets had to be modified . to study the statistical mechanics of the hamiltonian ( [ ham ] ) on sierpinski gaskets we have used monte carlo ( mc ) simulations based on metropolis algorithm @xcite .
for a gasket of given order the simulation would start at a low temperature with all phases aligned .
the first @xmath23 mc steps per site ( sps ) were thrown away , followed by seven mc links of @xmath23 mc sps each .
the values of @xmath23 which we used were 120000 , 360000 , and 600000 , depending on the system size and statistical errors .
at each temperature the range over which each angle @xmath24 was allowed to vary @xcite was adjusted to ensure an mc acceptance rate of about 50% .
the errors were calculated by breaking up each link into six blocks of @xmath23/6 sps , then calculating the average values for each of 42 blocks and finally taking the standard deviation @xmath25 of these 42 average values as an estimate of the error .
the final configuration of the angles @xmath26 at a given temperature was used as a starting configuration for the next higher temperature .
two types of boundary conditions were considered : closed , where the three corners of an @xmath20th order gasket were considered to be coupled to each other ( then each site has four nearest neighbors ) and open , where the three corners are uncoupled to each other . in figure 2
we show the results for the heat capacity per site calculated from the fluctuation theorem @xmath27 where @xmath28 is the boltzmann constant , @xmath29 is the absolute temperature and @xmath30 denotes the mc average .
the results for the four sierpinski gaskets were obtained with the closed boundary conditions . in the case of heat capacity
the results do not depend on boundary conditions as illustrated in figure 3 . for comparison ,
we have included in figure 2 the results obtained for the hamiltonian ( [ ham ] ) on three square lattices with the periodic boundary conditions .
the sizes of the square lattices were chosen so that the number of sites is similar to the number of sites for the three smaller sierpinski gaskets .
as there is no long - range order in two dimensions the heat capacity per site of the square lattices saturates with increasing system size @xcite . in the case of sierpinski gaskets
there is virtually no size dependence of the heat capacity per site .
this , however , does not mean that the model ( [ ham ] ) on two - dimensional sg also leads to the berezinskii - kosterlitz - thouless ( bkt ) transition resulting from vortex - antivortex unbinding at transition temperature @xmath31 .
the peak in the heat capacity of the square lattices is above the bkt transition temperature where the heat capacity has an unobservable essential singularity @xcite .
the peak is caused by unbinding of vortex clusters @xcite with increasing temperature above @xmath31 . in the case of sierpinski gaskets
the broad peak in @xmath32 could result from the average energy per site @xmath33 changing monotonically from the values near -2@xmath34 ( each site has four nearest neighbors ) at low temperatures to near zero at high temperatures ( disordered paramagnetic phase )
. a much better indicator of bkt transition is temperature dependence of the helicity modulus .
it measures the stiffness of the angles @xmath35 with respect to a twist at the boundary of the system .
when the hamiltonian ( [ ham ] ) is used to model a set of identical superconducting grains coupled via josephson tunneling , in which case @xmath24 is the phase of the superconducting order parameter on grain @xmath16 , the effect of applied magnetic field described by the vector potential @xmath36 is included via shift in the phase difference @xmath37 by @xmath38 where the line integral is taken along the line joining sites @xmath16 and @xmath39 and @xmath40 is the flux quantum @xcite .
for a _ uniform _ vector potential @xmath36 one has @xmath41 .
then , the helicity modulus can be defined @xcite as @xmath42 where @xmath43 is the helmholtz free energy . using @xmath44 one finds an expression for the helicity modulus which is analogous to ( [ cv ] ) . for the sierpinski gaskets @xmath36
was taken to be parallel to one of the sides of the triangles ( figure 1 ) and for the square latices @xmath36 was taken to be parallel to one of the sides of the squares .
our mc results for @xmath45 obtained with closed boundary condition for the sierpinski gaskets and with periodic boundary conditions for the square lattices are shown in figure 4 .
one expects a finite @xmath45 at low temperature when the angles are in an ordered configuration and zero stiffness in high temperature paramagnetic phase .
nelson and kosterlitz @xcite predicted a discontinuous jump in @xmath45 at the bkt transition temperature @xmath31 with a universal value @xmath46=2/@xmath15 , and the straight line in figure 4 gives the value of the jump at different temperatures .
for a finite system the jump in @xmath45 is replaced by continuous decrease with increasing @xmath29 which becomes steeper near the ( putative ) @xmath31 with increasing system size , as indicated by the results for the square lattices in figure 4 .
the two important aspects of our results are : ( 1 ) for all lattices considered a rapid downturn in @xmath45 starts near the universal 2/@xmath15-line .
( 2 ) while the low temperature values of @xmath45 for the square lattices do not depend on the system size , for the sierpinski gaskets they clearly decrease with increasing system size and the onset of the downturn in @xmath45 , which is in the vicinity of the universal 2/@xmath15-line , moves to the lower temperatures .
this suggests that in the thermodynamic limit ( i.e. when the order of the sg @xmath47 ) the angle stiffness @xmath45 would be zero at any finite temperature implying no bkt transition at a finite temperature .
we have examined the effect of boundary conditions ( closed vs. open boundary conditions ) and the results for the sixth order sg are shown in figure 5 . while the boundary conditions had no effect on the specific heat ( see figure 3 ) their effect on the helicity modulus is dramatic .
the open boundary conditions give zero helicity modulus at all temperatures within the error bars , which are much larger when the three corners of the sixth order sg are not coupled to each other .
such a dramatic change in @xmath45 implies that the closed boundary condition introduces significant additional correlations in two - dimensional sg fractal structure .
results obtained with the open boundary condition reinforce our conclusion about the absence of phase transition at any finite @xmath29 .
finally , for completeness , in figure 6 we show our results for the linear susceptibility per site computed from @xmath48 where @xmath49 is the magnetization of the system when hamiltonian ( [ ham ] ) is used to describe magnetic systems .
for bkt transition kosterlitz predicted @xcite that above @xmath31 the susceptibility diverges as @xmath50 $ ] , with @xmath51=0.25 , @xmath521.5 and @xmath53=0.5 , and is infinite below @xmath31 . for finite systems
one gets finite peaks in @xmath54 above @xmath31 which increase in height and move to lower temperatures with increasing system size , as illustrated by our results for three square lattices in figure 6 .
this trend is analogous to what one finds for ( [ ham ] ) in three dimensions where there _ is _ long - range order below the transition temperature .
therefore the simulation results for susceptibility are not useful in diagnosing bkt transition unless one has the results for very large cluster size and attempts to fit them according to prediction by kosterlitz @xcite .
we note that the susceptibilities for sierpinski gaskets obtained with closed boundary condition are considerably higher than those for the square lattices of comparable sizes .
also , the error bars are considerably larger than what was obtained for the square lattices .
the effect of boundary conditions on linear susceptibility for sierpinski gaskets is shown in figure 7 .
the main difference is that open boundary condition leads to much larger error bars at low temperatures which is consistent with our earlier observation that the closed boundary condition introduces additional correlations in the system .
our monte carlo results for the specific heat for the xy - model on two - dimensional sierpinski gaskets show that there is no long range order at any finite temperature as the specific heat is virtually independent of the system size .
moreover , the cluster - size dependence of the low temperature helicity modulus clearly indicates that there is no berezinski - kosterlitz - thouless transition in this planar fractal structure in the thermodynamic limit .
the low temperature values of the helicity modulus @xmath45 decrease with increasing system size while they are size - independent for square lattices where the bkt transition does take place . moreover , in the case of sierpinski gaskets the onset of the downturn in helicity modulus near the universal 2/@xmath15-line moves to the lower temperatures with increasing size of the clusters while for the square lattices only the rate of decline in @xmath45 beyond the 2/@xmath15-line increases with the system size approaching a discontinuous jump in the thermodynamic limit .
our results are consistent with previous findings of gefen et al .
@xcite who suggested , based on their exact renormalization group treatment of the resistor network on sierpinski gasket , that spin systems with continuous symmetry on this structure do not display a finite temperature phase transition .
also , our findings are consistent with the results of vallat et al .
@xcite obtained for the quadratic approximation to the xy - model on two - dimensional sierpinski gasket , who found that the vortex energy is finite and hence free vortices are present at any temperature .
this work was supported in part by the natural sciences and engineering research council of canada .
after this manuscript was accepted for publication it came to our attention that the possibility of phase transitions on lattices of effectively nonintegral dimensionality was first examined by dhar ( d. dhar , _ lattices of effectively nonintegral dimensionality _
, j. math .
. 18 ( 1977),pp .
577585 ) . in the case of the classical xy - model on the truncated tetrahedron lattice ( the effective dimensionality 2@xmath552/@xmath555 ) no phase transition at any finite temperature was obtained .
j.m gordon , a.m. goldman , j. maps , d. costello , r. tiberio , and b. whitehead , _ superconducting - normal phase boundary of a fractal network in a magnetic field _ ,
lett . 56 ( 1986 ) , | we have performed a monte carlo study of the classical xy - model on two - dimensional sierpinski gaskets of several cluster sizes . from the dependence of the helicity modulus on the cluster size
we conclude that there is no phase transition in this system at a finite temperature .
this is in agreement with previous findings for the harmonic approximation to the xy - model on sierpinski gasket and is analogous to the absence of finite temperature phase transition for the ising model on fractals with a finite order of ramification .
xy - model ; fractals ; sierpinski gasket ; monte carlo simulations |
spin - wave stiffness , @xmath3 , is an important parameter of magnetic materials that characterizes the magnon dispersion law , @xmath4 . in the mean - field approximation
it is directly related to exchange integral @xmath5 , namely , @xmath6 where @xmath7 is the gyromagnetic ratio , @xmath8 is the lattice constant , and @xmath9 is the spin.@xcite inelastic neutron scattering studies of the spin - wave stiffness in manganite single crystals @xcite revealed important information on the nature of their ferromagnetic transition . since this issue is under debate @xcite it is important to study spin - wave stiffness more closely , in particular , to measure @xmath3 in thin films .
it is also important to explore complementary techniques . in principle
, spin - wave excitations can be measured using magneto - optical kerr effect or brillouin light scattering @xcite although existing studies of manganites by these methods @xcite do not focus on spin waves .
spin - wave stiffness has been traditionally measured by the microwave absorption technique : using a fixed frequency esr spectrometer @xcite or broadband techniques.@xcite several groups reported fixed frequency microwave studies of standing spin - wave resonances ( swr ) in la@xmath10ba@xmath11mno@xmath2 ( ref .
@xcite ) , and la@xmath0mn@xmath12o@xmath2 ( refs .
@xcite ) thin films .
although spin - wave excitation at microwave frequencies was observed in la@xmath13sr@xmath14ca@xmath15mno@xmath2 ( ref .
@xcite ) , la@xmath0ca@xmath1mno@xmath2 ( ref .
@xcite ) and la@xmath0sr@xmath1mno@xmath2 ( ref .
@xcite ) thin films , it was not studied in detail .
in contrast to inelastic neutron scattering that ( i ) measures travelling spin waves with large wavevectors , @xmath16 1 - 0.1@xmath17 and ( ii ) requires large single crystals ; the microwave absorption technique measures spectrum of standing spin - wave resonances with small wavevectors , @xmath18@xmath17 and operates mostly with thin films .
the recent overview of spin - wave excitations in manganites @xcite compares thin - film microwave absorption measurements to inelastic neutron scattering studies of single crystals .
it turns out that there is no manganite compound that was studied simultaneously by both techniques .
our present work fills this gap .
we study spin - wave spectrum in epitaxial la@xmath0sr@xmath1mno@xmath2 films of different thicknesses and on different substrates by the microwave absorption technique at 9.4 ghz and compare our data to the inelastic neutron scattering studies on single crystals of the same composition .
consider a thin ferromagnetic film with an `` easy - plane '' magnetic anisotropy .
magnetic field is oriented at oblique angle @xmath19 with respect to the film normal .
orientation of magnetization , @xmath20 , is determined by the interplay between the external field , @xmath21 , and the perpendicular anisotropy field , @xmath22 , and is found from the following equation:@xcite @xmath23 where in - plane anisotropy has been neglected . in the presence of the microwave magnetic field with frequency @xmath24 , whereas @xmath25 , the spin - wave resonances ( swr ) are excited .
the resonance field @xmath21 is found from the following condition:@xcite @xmath26\nonumber\\ \times[\gamma h\cos(\theta-\psi)-\gamma h_{a}\cos2\theta+dk^{2 } ] \label{spin - wave}\end{aligned}\ ] ] if surface spins are completely pinned or completely unpinned , then @xmath27 where @xmath28 is the film thickness and @xmath29 is an integer .
the uniform fmr mode corresponds to @xmath30 , while the spin - wave resonances correspond to @xmath310 . for the perpendicular orientation ( @xmath32 )
( [ spin - wave ] ) yields @xmath33 where @xmath34 and @xmath24 is the microwave frequency . to determine @xmath3 ,
one measures microwave absorption in dependence of magnetic field and notices a sequence of resonances . by analyzing their spectrum using eq .
( [ sw ] ) one identifies mode numbers @xmath29 .
the slope of the linear dependence , @xmath35 vs @xmath36 , yields @xmath3 , while the intercept with the @xmath21-axis yields @xmath22 . to enable such procedure
the film thickness should lie in certain limits , @xmath37^{1/2 } \label{thickness}\ ] ] where @xmath38 is the skin - depth .
indeed , for efficient excitation of the spin - wave resonances the film thickness should be smaller than the skin - depth
. on the other hand , the film should be thick enough to support several resonances .
our experiments were performed with a bipolar @xmath39-band bruker esr spectrometer , a @xmath40 resonant cavity , and an oxford cryostat .
we studied la@xmath0sr@xmath1mno@xmath2 films ( @xmath41 50 , 100 , 150 and 200 nm ) grown on the ( 001 ) srtio@xmath2 substrate ( sto ) , and la@xmath10sr@xmath11mno@xmath2 films ( @xmath41 50 and 150 nm ) on the ndgao@xmath2 substrate ( ngo ) .
the samples were fabricated by the pulsed laser deposition @xcite and were cut to small mm - size pieces in order to keep the reasonable value of the cavity @xmath42-factor .
we measured magnetization and resistivity of these films by squid magnetometry and four - point technique , correspondingly .
the skin - depth at 9.4 ghz estimated from our resistivity measurements @xcite - 22 @xmath43 m at 295 k and 5 @xmath43 m at 50 k- considerably exceeds the film thickness .
the film uniformity could be estimated from the fmr spectrum .
a uniform film is characterized by a narrow fmr peak , corresponding to a well - defined anisotropy field , while a nonuniform film usually exhibits a broad fmr peak indicating wide spread of the anisotropy fields .
most part of our samples demonstrated narrow fmr peaks at ambient temperature , although some samples showed several narrow fmr peaks corresponding to discrete values of anisotropy field .
figure [ fig : lsmo55-h ] shows microwave absorption derivative in the perpendicular field for a 200 nm thick la@xmath0sr@xmath1mno@xmath2 film on the srtio@xmath2 substrate .
we observe a series of slightly asymmetric narrow peaks .
the asymmetry arises from the coupling to the dielectric resonances in the sto substrate and was observed in other studies as well .
@xcite the angular dependence of the resonant fields ( not shown here ) follows eq .
( [ sw ] ) , hence we attribute these peaks to the spin - wave resonances .
we cut the original 5@xmath44 5 mm@xmath45 film to five pieces with the size of 1@xmath44 0.5 mm@xmath45 and all these pieces demonstrated fairly identical spectra .
the very fact that we observe sharp and reproducible resonances proves that the film is highly uniform .
another indication of the high quality of the film is the narrow peak - to - peak linewidth ( 16 oe at ambient temperature ) and low coercive field ( 11 - 16 oe at 295 k ) .
the mode numbers were established as follows .
we assigned consecutive numbers to the peaks in fig .
[ fig : lsmo55-h ] and checked whether linear dependence , @xmath46 , predicted by eq .
( [ sw ] ) , holds .
the best correspondence to eq .
( [ spin - wave ] ) was achieved for the sequence @xmath471,2,3 .. or @xmath470,2,3 ... there is some ambiguity in whether the strongest peak corresponds to @xmath30 , to @xmath48 or to their sum , since the splitting between the @xmath30 and the @xmath48 modes as predicted by eq .
( [ sw ] ) , is only 25 oe and this is comparable to the linewidth .
figure [ fig : linewidth - k ] shows swr intensities and linewidths .
the linewidth increases almost linearly with mode number , while intensity decreases .
odd modes have generally higher intensities than even modes ( see also fig . [
fig : lsmo55-h ] ) .
figure [ fig : lsmo55-harm ] shows dependence of the resonance fields on the mode number .
the higher - order modes obey quadratic dependence , @xmath49 , while the modes with low @xmath29 show tendency to linear spacing that is quite common for the films with surface pinning .
@xcite we exclude from our analysis the first two modes that should be most strongly affected by surface pinning and determine spin - wave stiffness from the slope of @xmath35 vs @xmath36 dependencies using eq .
( [ sw ] ) .
the results are shown in fig .
[ fig : d - total ] . to find the perpendicular anisotropy field
we extrapolate the @xmath35 vs @xmath50 dependencies to @xmath30 .
equation ( [ anisotropy ] ) and fig .
[ fig : lsmo55-harm ] yield @xmath510.4 t at 295 k and @xmath511 t at 4.2 k. to analyze these values we note that the perpendicular anisotropy field of a thin film , @xmath52 consists of demagnetizing field- @xmath53 , crystalline anisotropy- @xmath54 , and stress - induced anisotropy- @xmath55 .
the crystalline anisotropy of la@xmath0sr@xmath1mno@xmath2 is very small @xcite and does not exceed 0.03 t.@xcite demagnetization field as estimated from magnetization @xcite is @xmath560.74 t at 4.2 k. since the lattice mismatch between la@xmath0sr@xmath1mno@xmath2 and srtio@xmath2 is only 1.4 @xmath57 and the film is sufficiently thick , the @xmath55 is not high and achieves considerable magnitude only at low temperatures.@xcite equation ( [ anisotropy ] ) yields the stress anisotropy field , @xmath58 0.26 t at 4.2 k. this means that even at low temperatures the demagnetization field is the dominant contribution to the anisotropy field .
this is consistent with other measurements . indeed ,
magnetization studies of the ref .
@xcite for the films of comparable thickness found that @xmath59 at 10 k , while ref .
@xcite found that @xmath60 at 295 k. figure [ fig : ngo - h ] shows microwave absorption spectrum for a 150 nm thick la@xmath10sr@xmath11mno@xmath2 film on the ndgao@xmath2 substrate at 108 k. we observe a strong and narrow peak at 10445 oe and a series of low - field satellites .
the peak - to - peak linewidth of the dominant resonance is very small ( 12 oe at ambient temperature and 33 oe at 108 k ) and this proves high quality of the film .
the coercive field at ambient temperature is only 4 oe . we cut the film to several pieces and they showed consistent spectra . to identify the resonances we measured temperature and angular dependencies of the resonant field and came to conclusion that the peaks designated with integer numbers in fig .
[ fig : ngo - h ] are spin - wave satellites of the peak at 10445 oe , while the strong peak at @xmath61 7470 oe does not belong to this series .
indeed , the angular dependencies of resonant field of the numbered peaks are very similar ( not shown here ) and different from that for the peak at 7470 oe .
the same is true with respect to the temperature dependencies .
we attribute the peak at 7470 oe to the region with a different discrete value of the anisotropy field and exclude it from the subsequent analysis .
figure [ fig : ngo - harm ] shows that the resonant fields of higher modes follow @xmath49 dependence .
the slope of this dependence ( the first mode excluded ) yields the spin - wave stiffness @xmath3 .
the results are plotted in fig .
[ fig : d - total ] .
the same data yield the perpendicular anisotropy field . we find @xmath510.73 t at 4.2 k. this is almost equal to the demagnetizing field , @xmath620.74 t , and means that the stress - induced anisotropy is negligible here , as expected for the lattice - matched substrate .
the linewidth ( see inset ) steadily increases with @xmath29 .
we observed spin - wave resonances in several films with different thickness . in a very thin ( @xmath41 50 nm ) la@xmath10sr@xmath11mno@xmath2 film on srtio@xmath2
we observed only one spin - wave resonance which was displaced down by 0.23 t from the dominant fmr resonance [ while eq .
( [ sw ] ) predicts 0.08 t ] .
such pronounced displacement is obviously related to strong surface pinning , hence it is not possible to measure @xmath3 there . in several 150 nm thick la@xmath10sr@xmath11mno@xmath2 films on srtio@xmath2 we observed two overlapping series of spin - wave resonances , so that determination of @xmath3 here was ambiguous .
@xcite observed such split spin - wave resonances in a 350 nm thick la@xmath0mn@xmath12o@xmath2 film on laalo@xmath2 substrate and showed that the splitting disappears after annealing in oxygen .
following ref .
@xcite we studied the effect of oxygen annealing on the swr spectra in our films .
contrary to ref .
@xcite we found that oxygen annealing at different temperatures from 600@xmath63c to 900@xmath63c does not ameliorate the swr spectrum in our films but introduces additional splitting .
in particular , the resonant peaks in annealed films split into many narrow lines with the spacing of @xmath6415 oe . the difference between our results and those of ref .
@xcite is probably related to the fact that we operate with the film of different composition and on different substrate .
figure [ fig : d - total ] compares the spin - wave stiffness found in our microwave absorption studies of la@xmath0sr@xmath1mn@xmath0o@xmath2 thin films to inelastic neutron scattering measurements on single crystals of the same composition.@xcite consider first the limit @xmath65 .
our measurements for two films of different thickness yield @xmath66210 mev @xmath45 while the corresponding neutron scattering data show lower values , @xmath67 188 ( ref .
@xcite ) , 170 ( ref .
@xcite ) , 176 ( ref .
@xcite ) , and 190 mev @xmath45 ( ref .
the discrepancy between @xmath68 and @xmath69 may arise from the fact that few atomic layers adjacent to the film - substrate interface are nonmagnetic . in lsmo
this `` dead layer '' may be up to 5 nm thick,@xcite hence `` magnetic '' thickness that appears in eq .
( [ sw ] ) is smaller than the nominal thickness .
the correction for the `` dead layer '' can bring down @xmath68 by @xmath70 . the spin - wave stiffness at @xmath65 can be also estimated from the temperature dependence of magnetization and the @xmath71-bloch law ( see also ref .
@xcite ) . for the la@xmath0sr@xmath1mno@xmath2 single crystals this yields @xmath72154 ( ref .
@xcite ) , while for ceramics of the same composition , @xmath72197 mev @xmath45 ( ref .
@xcite ) . consider now the temperature dependence of the spin - wave stiffness .
figure [ fig : d - total ] shows that the results for two films with different thicknesses and on different substrates ( one with tensile stress - sto and another with a weak compressive stress - ngo ) are very close , as expected for intrinsic property . at ambient temperature
, we find for these two films @xmath73104 and 114 mev @xmath45 , correspondingly .
this is almost identical to the single crystal data at ambient temperature- @xmath74114 ( ref .
@xcite ) and 100 mev @xmath45 ( ref .
however , one should take into account the difference in @xmath75 as well . in order to compare data for the samples with different @xmath75 we consider @xmath76 vs @xmath77 plots where the temperature is an implicit variable .
a similar plot was used earlier by ref .
@xcite to compare spin - wave stiffness of the fe - cr alloys with different composition and different @xmath75 .
the rationale behind such plot is the mean - field expression -eq .
( [ exchange])- that in fact relates the spin - wave stiffness to magnetization , @xmath78 . here
, @xmath79 is the bohr magneton , @xmath80 is the @xmath80-factor , and @xmath81 is the atomic volume . equation ( [ exchange ] ) yields direct proportionality , @xmath82 . strictly speaking
, this proportionality is expected only for the heisenberg model and at @xmath830 . however , since @xmath84 and @xmath3 both go to zero at @xmath75 , while @xmath5 , in general , varies continuously across the ferromagnetic transition ; then we expect that the @xmath85 proportionality holds up to @xmath75 .
indeed , as the ref .
@xcite shows , empirical data on many ferromagnetic compounds suggest that @xmath3 and @xmath84 have the same temperature dependence in the whole range from @xmath83 0 to @xmath86 . to effectuate this approach we plot @xmath3 vs @xmath84 ( fig .
[ fig : m - d - all ] ) where @xmath3 and @xmath84 are measured at the same temperature.@xcite we find @xmath3 from the microwave absorption spectra , while the magnetization is estimated indirectly , from the anisotropy field , @xmath22 . indeed , since the dominant contribution to @xmath22 in our films is the demagnetizing field @xmath53 ( especially for the films on ngo ) , hence the anisotropy field is a measure of magnetization .
figure [ fig : m - d - all ] plots the spin - wave stiffness versus anisotropy field where the temperature is an implicit parameter .
it also plots the corresponding single crystal neutron - scattering data of ref .
@xcite where magnetization was estimated from the intensity of the electronic bragg peak .
the upper horizontal scale in fig .
[ fig : m - d - all ] was chosen in such a way that the low - temperature limit of the electronic bragg peak intensity ( @xmath87 ) corresponds to saturation magnetization of la@xmath0sr@xmath1mno@xmath2 , i.e. @xmath62 0.74 t. the horizontal error bars in our thin film measurements take into account the possible difference between the anisotropy field and magnetization arising from the stress anisotropy field [ eq .
( [ anisotropy ] ) ] .
we assume extreme values of @xmath88=-0.06 and 0.2 for the films on ngo and sto , correspondingly .
we observe that above @xmath83295 k the data for both our films and for the single crystal collapse .
the resulting @xmath89 dependence is quaisilinear that indicates the same critical indices of @xmath3 and @xmath84 at @xmath75 .
this is not obvious since the neutron - scattering measurement were performed in zero magnetic field while the microwave measurements were performed in finite field of 0.4 t to 1 t. using eq .
( [ exchange ] ) we find the same value of the exchange integral , @xmath903.6 mev ( for @xmath911.85 and @xmath92 ) for the film and single crystal .
this is also not obvious since thin films are strained .
we processed in the same way the microwave absorption data of ref .
@xcite for the film of different composition -la@xmath0mn@xmath12o@xmath2 ( see inset in fig . [
fig : m - d - all ] ) .
while @xmath89 proportionality holds at low temperatures , there is a an upward deviation as @xmath93 .
this suggests discontinuous variation of @xmath3 across the ferromagnetic transition . our data for the la@xmath0sr@xmath1mno@xmath2 films do not indicate such discontinuity .
the linewidth that steadily increases with mode number ( figs .
[ fig : linewidth - k],[fig : ngo - harm ] ) is not frequently observed in microwave absorption studies of spin - wave resonances .
previous studies of permalloy films and other highly conducting ferromagnets found weak and nonmonotonous dependence @xmath94 .
@xcite ) such nonmonotonous dependence arises from the sum of several sources such as eddy - current damping , surface roughness , and fluctuations ( exchange energy , anisotropy field , thickness ) .
@xcite the @xmath95-dependent linewidth observed in our studies could be hardly intrinsic . indeed , theoretical prediction for the magnon - magnon scattering in manganites yields inverse spin lifetime @xmath96 ( ref .
the neutron scattering studies of la@xmath97sr@xmath98mno@xmath2 single crystals @xcite indeed yield @xmath99 ( here @xmath95 is in @xmath17 and @xmath100 is in mev ) .
extrapolation of this @xmath101 dependence ( obtained for 10 k and @xmath1020.02 - 0.16@xmath17 ) down to the range of wavevectors occurring in our microwave studies ( @xmath103 ) yields @xmath104 .
this is much smaller than what we observe in fig .
[ fig : linewidth - k ] - @xmath105 ( note however that our films have different composition as compared to those studied in ref .
this indicates that the @xmath95- dependent linewidth found in our studies in thin films is extrinsic .
it most probably arises either from the inhomogeneous broadening associated with thickness nonuniformity , @xcite or from the spin - wave scattering on surface and bulk disorder .
@xcite if we take an extreme approach and assume only thickness nonuniformity , then ref .
@xcite yields @xmath106 where @xmath107 is the linewidth of the uniform precession mode and @xmath108 is the average thickness variation .
analysis of our data according to eq .
( [ linewidth - thickness ] ) ( dashed lines in figs .
[ fig : linewidth - k],[fig : ngo - harm ] ) yields @xmath108= 4 nm ( 2@xmath57 ) and 6 nm ( 4@xmath57 ) for the films on sto and on ngo , correspondingly .
if we go to another extreme and assume spin - wave scattering to be the only source of line broadening , then according to the transit - time treatment @xcite we find @xmath109 where @xmath110 is the group velocity and @xmath111 is the mean free path .
spin - wave scattering on bulk disorder is characterized by the combination of linear and quadratic @xmath95-dependence , @xcite whereas the linear @xmath95-dependence implies constant mean free path .
analysis of our data according to eq .
( [ sc ] ) and assuming constant @xmath111 , yields @xmath112 2 for the lsmo / sto film at ambient temperature ( fig .
[ fig : linewidth - k ] ) . for the lsmo / ngo film ( fig .
[ fig : ngo - harm ] ) we find @xmath112 1.4 and 1.3 at 108 k and at 187 k , correspondingly .
the spin - wave mean - free - path being on the order of film thickness , suggests that the scattering occurs predominantly at film interfaces .
such scattering may result from surface roughness @xcite or from the localized mn@xmath113-ions @xcite or mn@xmath114-ions @xcite at film surface .
the observed @xmath95-dependence of the linewidth is most probably due to both mechanisms : thickness nonuniformity and scattering .
note however , that both extreme approaches [ eq .
( [ linewidth - thickness ] ) and eq .
( [ sc ] ) ] yield long mean free path that practically excludes bulk scattering .
this is surprising , since the stoichiometry of la@xmath0sr@xmath1mno@xmath2 suggests two possible states of mn - ion : mn@xmath115 with spin @xmath9=3/2 and mn@xmath113 with spin @xmath912 .
if la@xmath0sr@xmath1mno@xmath2 were heisenberg ferromagnet this would certainly result in magnetic inhomogeneity .
however , observation of standing spin - wave resonances with high mode numbers indicates negligible bulk scattering and a single magnetic state of mn - ions in the bulk , as was already established by the nmr studies of mn@xmath116 in la@xmath10sr@xmath117mno@xmath2 ( ref .
@xcite ) .
the standing spin - wave resonances have been observed so far in the manganite compounds of the general formula la@xmath118a@xmath119mno@xmath2 where a is sr , ca , ba , mn and @xmath120 ( refs . @xcite and present work ) .
the @xmath120 composition corresponds to highest conductivity and is the most distant from phase boundaries , @xcite in other words , the phase separation effects should be least insignificant here .
this composition has also the highest spin - wave stiffness @xcite that indicates increased length of the exchange interaction . the very fact that so far there is no indication of the standing spin - wave resonances in manganite compounds with @xmath121 ( we also did nt find any spin - wave resonances in la@xmath122sr@xmath123mno@xmath2 films ) may serve as an indirect evidence of strong spin - wave scattering there .
we measured temperature dependence of the spin - wave stiffness in thin la@xmath0sr@xmath1mno@xmath2 films as determined from the standing spin - wave resonances at microwave frequencies . at ambient temperature ,
the spin - wave stiffness of thin films and of single crystals is the same , while at low temperatures the spin - wave stiffness of thin films is enhanced with respect to that of a single crystal .
the spin - wave linewidth in our films is limited by the scattering at film interfaces .
the very fact that we are able to observe spin - wave resonance in la@xmath0sr@xmath1mno@xmath2 up to eighth order implies high degree of coherence and very low bulk spin - wave scattering .
this is important for spintronics and means that la@xmath0sr@xmath1mno@xmath2 can be used as a spin conserving component .
we are grateful to denis golosov for illuminating discussions , to xiangzhen xu for the help with the handling of the samples , to oscar arnache who initiated these experiments , and to anonymous referee for instructive comments . c. kittel , phys .
* 110 * , 1295 ( 1958 ) .
lynn , r.w .
erwin , j.a .
borchers , q. huang , a.a .
santoro , j.l .
peng , z.y .
li , phys .
lett . * 76 * , 4046 ( 1996 ) .
martin , g. shirane , y. endoh , k. hirota , y. moritomo , y. tokura , phys .
b * 53 * , 14285 ( 1996 ) .
l. vasiliu - doloc , j.w .
lynn , y.m .
mukovskii , a.a .
arsenov , and d.a .
shulyatev , j. appl .
* 83 * , 7342 ( 1998 ) .
moudden , l. vasiliu - doloc , l. pinsard , a. revcolevschi , physica b * 241 * , 276 ( 1998 ) .
l. vasiliu - doloc , j. w. lynn , a. h. moudden , a. m. de leon - guevara , and a. revcolevschi , phys .
b * 58 * , 14913 ( 1998 ) .
j. zhang , f. ye , hao sha , p. dai , j. a. fernandez - baca , and e.w.plummer , j. phys . :
matter * 19 * , 315204 ( 2007 ) .
m.b . salamon and m. jaime , rev .
phys . * 73 * , 583 ( 2001 ) . c. e. patton , phys
. rep . * 103 * , 251 ( 1984 ) .
f.l.a . machado , m.a .
lucena , a.e.p .
de araujo , a. azevedo , f.m
. de aguiar , s.m .
rezende , a.k .
nigam , physica b * 320 * , 119 ( 2002 ) .
d. talbayev , h. zhao , g. lupke , j. chen , qi li , appl .
* 86 * , 182501 ( 2005 ) .
m. farle , rep .
. phys . * 61 * , 755 ( 1998 ) .
v.p . denysenkov , a.m. grishin , rev .
instrum . * 74 * , 3400 ( 2003 ) .
lofland , s.m .
bhagat , c.kwon , m.c.robson , r.p .
sharma , r. ramesh , t. venkatesan , phys .
a * 209 * , 246 ( 1995 ) .
v. dyakonov , a. prohorov , v. shapovalov , v. krivoruchko , v. pashchenko , e. zubov , v. mihailov , p. aleshkevych , m. berkowski , s. piechota , h. szymczak , j. phys.:cond . matter * 13 * , 4049 ( 2001 ) .
v. dyakonov , a. prohorov , v. shapovalov , s. khartsev , v. krivoruchko , v. mihailov , v. pashchenko , e. zubov , p. aleshkevych , k. dyakonov , s. piechota , h. szymczak , phys .
lett . a * 268 * , 202 ( 2000 ) .
p.aleshkevych , m.baran , v. dyakonov , r. szymczak , h. szymczak , k. baberschke , j. lindner , k. lenz , phys .
( a ) * 203 * , 1586 ( 2006 ) .
j. yin , y.x .
sui , j.h .
zhang , x.y .
liu , z.g .
liu , phys .
sol . ( a )
* 174 * , 499 ( 1999 ) .
a.i . shames , e. rozenberg , g. gorodetsky , m. ziese , in : _ nanostructured magnetic materials and their apllications _ , pp.205 - 212 , kluwer acad .
lyfar , s.m .
ryabchenko , v.n .
krivoruchko , s.i .
khartsev , a.m. grishin , phys .
b * 69 * , 100409(r ) ( 2004 ) .
p.aleshkevych , m.baran , and h.szymczak , acta physica polonica * 106 * , 593 ( 2004 ) .
a.g . gurevich and g.a .
melkov , magnetization oscillations and waves , crc press , boca raton , 1996 , p.19 .
m. golosovsky , m.abu-teir , d. davidov , o. arnache , p. monod , n. bontemps , r.c.budhani , j. appl . phys . * 98 * , 084902 ( 2005 ) .
k. senapati and r. c. budhani , phys .
b * 71 * , 224507 ( 2005 )
phillips , proc .
royal soc .
a * 292 * , issue 1429 , p.224 ( 1966 ) .
y. suzuki , h.y .
hwang , s.w .
cheong , r.b .
van dover , appl .
. lett . * 71 * , 140 ( 1997 ) .
m. ziese , h.c .
semmelhack , p. busch , j. magn .
magn . mat .
* 246 * , 327 ( 2002 ) .
lofland , s.m .
bhagat , c.kwon , s.d .
tyagi , y.m .
mukovskii , s.g .
karabashev , a.m. balbashov , j. appl . phys . * 81 * , 5737 ( 1997 ) .
l. ranno , a. llobet , r. tiron , e. favre - nicolin , appl .
science * 188 * , 170 ( 2002 ) .
b. wiedenhorst , c. hofener , yafeng lu , j. klein , l. alff , r. gross , b.h .
freitag , w. mader , appl .
* 74 * , 3636 ( 1999 ) .
sidorenko , g.allodi , r. de renzi , g. balestrino , m.angeloni , phys .
b * 73 * , 054406 ( 2006 ) .
n. rosov , j.w .
lynn , j. kastner , e.f .
wassermann , t. chattopadhyay , h. bach , j. appl . phys . * 75 * , 6072 ( 1994 ) .
smolyaninova , j.j .
hamilton , r.l .
greene , y.m .
mukovskii , s.g .
karabashev , a.m. balbashov , phys .
b * 55 * , 5640 ( 1997 ) .
s. budak , m. ozdemir , b. aktas , physica b * 339 * , 45 ( 2003 ) .
t. rossing and t.g .
phillips , j. appl .
phys . * 39 * , 1375 ( 1968 ) .
u. kobler , a. hoser , w. schafer , physica b-cond.matt .
* 364 * , 55 ( 2005 ) .
our measurements were performed with the thermocouple mounted at some distance from the sample . since thermalization is slow , to restrict the duration of the experiment to several hours , in some experimental runs we slowly heated the sample from low temperature and measured the spectra in a continuous way without waiting for thermalization .
in such experiments we determine only @xmath3 and @xmath22 .
p. lubitz , s.m .
bhagat , g.c .
bailey , c. vittoria , phys .
b * 11 * , 3585 ( 1975 ) .
g. suran and r.j .
gambino , j. appl .
phys . * 50 * , 7671 ( 1979 ) . f. schreiber and z. frait , phys . rev .
b * 54 * , 6473 ( 1996 )
bailey and c. vittoria , phys .
* 28 * , 100 ( 1972 ) .
phillips and h.m .
rosenberg , rep .
* 29 * , 285 ( 1966 ) .
levy , phys .
b * 25 * , 2893 ( 1982 ) .
golosov , phys .
lett . * 84 * , 3974 ( 2000 ) .
y. motome and n. furukawa , phys .
b * 71 * , 014446 ( 2005 ) .
u. hoeppe and h. benner , phys .
b * 71 * , 144403 ( 2005 ) .
scotter , j. phys .
phys . * 5 * , l93 ( 1972 ) .
a. azevedo , a. b. oliveira , f. m. de aguiar , and s. m. rezende , phys .
b * 62 * , 5331 ( 2000 ) .
de jong , i. bergenti , v.a .
dediu , m. fahlman , m. marsi , c. taliani , phys .
b * 71 * , 014434 ( 2005 ) .
m. imada , a.fujimori , y.tokura , rev . mod
. phys . * 70 * , 1039 ( 1998 ) . | we studied magnetic field dependent microwave absorption in epitaxial la@xmath0sr@xmath1mno@xmath2 films using an x - band bruker esr spectrometer . by analyzing angular and temperature dependence of the ferromagnetic and spin - wave resonances we determine spin - wave stiffness and anisotropy field .
the spin - wave stiffness as found from the spectrum of the standing spin - wave resonances in thin films is in fair agreement with the results of inelastic neutron scattering studies on a single crystal of the same composition [ vasiliu - doloc et al .
, j. appl .
phys . * 83 * , 7343 ( 1998 ) ] . |
this paper expands the concept of generalised cesaro convergence introduced in [ 1 ] and uses this to introduce a notion of generalised root identities . in turn
these identities are used to investigate both the asymptotic behaviour and the geometric distribution of roots and poles of the gamma function , @xmath2 , and the riemann zeta function , @xmath3 . in section 2.1
we introduce the core notion of `` remainder '' cesaro summation / convergence , where the partial - sum function analysed via a cesaro scheme is now on a contour relative to chosen point @xmath7 , rather than always being on @xmath8 . as an initial example
we show that the hurewicz zeta function , @xmath9 , is defined naturally in this context , and that well - known relationships among bernoulli polynomials follow naturally within this remainder cesaro approach . in section 2.2 this
is extended in the usual way to a notion of remainder cesaro products and , when applied to the elementary case of @xmath10 , is used to give a cesaro definition of the gamma function , @xmath2 , at least after ensuring that the geometric location of terms in the remainder sum / product is respected in the cesaro analysis . in section 2.3
we demonstrate the utility of this remainder cesaro approach , and in particular its geometric intuitiveness when coupled with the dilation - invariance of cesaro convergence , in analysing three example results .
the first is an elementary derivation of the duplication formulae for @xmath2 ; the second is the derivation of the functional equations for @xmath2 and @xmath3 , by exploiting symmetries in the cesaro context to introduce bi - directional sums which are apriori periodic and thus amenable to fourier series analysis ; and the third is a short derivation of an interesting integral identity for @xmath9 . in section 3
we then turn to the question of root identities , and in particular the possibility of generalising the well - known identities relating roots and coefficients of polynomials to some more general class of functions . in section 3.1 a family of such identities , for arbitrary @xmath11 and @xmath12 ,
are initially derived on a formal , heuristic basis . in general , these equate a quantity defined in terms of the derivatives of a function ( the `` derivative side '' ) with a quantity defined in terms of the roots of the function ( the `` root side '' ) .
the latter quantity in fact includes equally contributions from roots ( with multiplicity @xmath13 ) , poles ( with @xmath14 ) and even branch points ( with e.g. @xmath15 ) of a function ; all of which cases are thus subsumed thereafter under the label of `` generalised roots '' .
a natural equivalence is then established between satisfying the root identities at a single @xmath16 , and satisfying them at arbitrary @xmath11 ( at least an open neighbourhood ) for @xmath17 alone .
unfortunately it is readily seen that most non - polynomial functions do not satisfy these generalised root identities .
however , at least for those where the discrepancy is sufficiently well - behaved ( as a function of @xmath7 ) , we show that it is possible to find an `` equivalent '' function with the same root - set which does satisfy these identities .
the construction of this equivalent function by iterative removal of obstructions to the successive ( @xmath18 ) root identities at a given fixed @xmath7 is outlined . in section 3.2
, however , we verify that in practice many natural and important functions do already satisfy these identities with no , or minimal , adjustment by considering three examples which will become primary focuses of attention in this article . in the first
we show that @xmath19 satisfies the identities for all @xmath11 and all @xmath12 and we note that this leads naturally to evaluation of some values of @xmath3 , for example euler s famous evaluation of @xmath20 . in the second
we verify that @xmath21 also satisfies the root identities for all @xmath11 and all @xmath12 , albeit in the case of @xmath17 that a renormalisation is required on the root side to handle uniformly the ln - divergences which arise for any @xmath7 and which do not have cesaro limits . in the third
we find that @xmath3 alone does not satisfy the root - identity for @xmath17 , but see by experimentation using the first @xmath22 non - trivial zeros that the obstruction in this case appears to be a constant , @xmath23 , independent of @xmath24 .
this suggests that instead @xmath25 does satisfy the @xmath17 root identity for all @xmath24 and we find that in fact this is true and reflects a well - known explicit formula for @xmath3 which represents the hadamard product formula for @xmath26 .
thus , in this case , the generalised root identity for @xmath17 is equivalent to the hadamard product formula ; for @xmath27 the extra factor @xmath28 no longer contributes on either the derivative or root sides and so @xmath3 itself then satisfies these identities for all @xmath24 directly . in section 3.3
we then turn to the possibility of further generalising our root identities by allowing @xmath0 to be not just a positive integer , but @xmath29 arbitrary , with the cases of @xmath1 being of particular interest .
we begin by discussing extensively how to even make sense of the identities in the case of arbitrary complex @xmath0 . on the root side a generalised convergence scheme - always cesaro in this
paper - is required to make sense of what are now usually divergent sums , and we find that it is crucial that the summands , @xmath30 , in this cesaro context be added in geometrically at the shifted points @xmath31 . on the derivative side the notion of @xmath32 now becomes problematic ; we note various basic properties it should have and then define it consistently with these via fourier theory , in a manner familiar from pseudodifferential operators . as a first example of the value of this generalisation to arbitrary complex @xmath0 we give
a short derivation showing that , for our example of @xmath19 , @xmath33 satisfying the generalised root identities for all @xmath7 and all @xmath29 is equivalent to the functional equation for @xmath3 . with this as indication of the utility of this extension , we then discuss these generalised identities in the particular cases of @xmath1 where the @xmath34 factor becomes singular . for the remainder of the paper ,
in fact , we shall ignore the case of arbitrary @xmath29 and focus exclusively on @xmath1 . in this case
we show how to interpret our derivative - side definition via a distributional interpretation of the expressions @xmath35 and note that in fact it is this singular behaviour that generally makes calculation of the derivative sides more tractable for @xmath1 than for arbitrary @xmath29 since it allows omission of hard - to - compute finite integral contributions , leaving only these distributional contributions .
we then discuss at some length why , for a function @xmath33 satisfying the generalised root identities , the identities for @xmath1 should naturally give us much more information than the corresponding cases of @xmath12 both about the asymptotic behaviour of @xmath33 and about the asymptotic and geometric distribution of its roots .
we conclude section 3.3 with a quick overview of the initial formal fourier theory results that we will use in the remainder of the paper , which is concerned with analysing the implications of the generalised root identities for our remaining two key examples , @xmath21 and @xmath36 .
in section 3.4 we consider the case of @xmath21 .
we first verify that @xmath21 does in fact continue to satisfy the root identities for all @xmath1 by explicit calculation of both the root sides ( using cesaro ) and derivative sides ( using fourier and distributional calculations ) in these cases .
then , as an example of the asymptotic information contained within the identities for @xmath1 we show how , for this example of @xmath2 , we can use them to deduce stirling s theorem describing the asymptotics of @xmath21 .
we do this in an engineering fashion by successively using the identities for @xmath37 to construct the next order in the stirling asymptotics .
we show , using the generating function which defines the bernoulli numbers , why the stirling formula in turn leaves the root identities for @xmath12 in tact , without need of further correction terms to remove obstructions in these cases .
finally , we conclude section 3.4 with some discussion of why our derivation of stirling s theorem in fact implies a `` two - sided '' result giving stirling - like asymptotics also as @xmath38 , in addition to the actual stirling asymptotics as @xmath39 . we use the functional equation for @xmath2 to verify that this is indeed the case , but with specific additional `` one - sided '' asymptotic contributions as @xmath38 , albeit ones which did not derail our derivation above . in section 4
we then turn to consideration of the generalised root identities for @xmath1 for @xmath3 . in section 4.1
we consider the derivative sides of these identities and demonstrate that in fact they are all identically zero for all @xmath24 when @xmath1 .
this is done by using the euler product formula for @xmath3 together with a basic property of @xmath40 to in fact derive an explicit formula for the derivative side as a sum over primes @xmath41 . in section 4.2 ,
which then takes up the bulk of the remainder of the paper , we consider the cases of @xmath42 , @xmath43 and @xmath6 in turn on the root sides with a view to seeing whether these root sides can likewise be seen to be identically zero for all @xmath24 in spite of not knowing the exact location of the non - trivial roots ; and indeed whether any asymptotic or geometric information can be gleaned on the location of these non - trivial zeros from the requirement that the root - sides be zero in these cases . in section 4.2.1
we consider the case of @xmath42 , in section 4.2.2 the case of @xmath43 , and in section 4.2.3 the case of @xmath6 . in all cases
the evaluation of the root - sides proceeds by considering first the trivial roots , t , then the simple pole and finally the non - trivial roots , nt . as expected , the majority of the effort is in the computations for the contributions from nt .
here we rely on the explicit riemann - von mangoldt formula for @xmath44 , the function which counts non - trivial roots with imaginary parts between @xmath45 and @xmath46 ; this expresses @xmath44 as the sum of three pieces @xmath47 where @xmath48 is the leading asymptotic part of @xmath44 , @xmath49 is the famous argument of the riemann zeta function , and @xmath50 is a completely explicit function giving the negative order asymptotics of @xmath44
. the calculations of the root - side contribution from nt always thus further subdivide into calculations of the cesaro limits for the terms arising from each of these three pieces . in carrying out these cesaro computations
we confirm that it is absolutely critical when @xmath1 both ( a ) to respect the location of the roots in determining the geometric placement of summands and ( b ) to independently carry out computations for the nt - roots above and below the real axis using separate parameters @xmath46 and @xmath51 ( and associated geometric complex variables @xmath52 and @xmath53 ) , but to recombine these into a 2-d partial sum function ( of @xmath52 and @xmath53 ) before taking limits .
failure to do so leads to incorrect evaluation and paradoxes involving apparent inconsistency between the results for @xmath42 and @xmath43 .
for @xmath42 in section 4.2.1 we verify that we do indeed have the root - side identically zero for all @xmath24 , at least after assigning cesaro limit @xmath45 to a residual 2-d log - term , @xmath54 ( with this latter seemingly ad hoc calculation either following from a suitable 2-d cesaro analysis , or alternatively to be treated as a consequence of the root identity for @xmath42 , allowing it at least to be used confidently for the subsequent cases of @xmath43 and @xmath6 where it also arises ) .
we return to address it again and justify why assigning this 2d cesaro limit the value @xmath45 is legitimate ] for @xmath43 in section 4.2.2 we likewise verify that we do indeed have the root - side identically zero for all @xmath24 , at least subject to an estimate on @xmath49 which is known to hold conditional on the riemann hypothesis ( rh ) .
for @xmath6 in section 4.2.3 , however , we find that a symmetry of the nt - roots is broken .
for @xmath42 and @xmath43 the fact that any non - trivial roots off the critical line occur in mirror pairs meant that the net contributions from any such pairs to our cesaro calculations was exactly as for a corresponding double - root on the critical line , and so no explicit terms expressing potential contributions from departures from the critical line arose .
but for @xmath6 there is such a term , @xmath55 , and after carrying out the analogous , albeit more intricate , computations for the root side in this case we find that the value of the root - side for @xmath6 is given by @xmath56 . in order for the @xmath6 root identity to hold
, this would need to be identically zero and this allows us to conclude section 4.2.3 by inferring , as a consequence , that the riemann hypothesis must be false . given the nature of this claim , we end by including a final section , section 4.3 , itemising and discussing the admittedly many areas where the arguments in the paper up to this point have been informal , heuristic or otherwise in need of more rigorous treatment .
we nonetheless note that , while our emphasis throughout the paper is on developing and applying calculational tools rather than on formal rigour , we believe the nature of the resulting computations is sufficiently suggestive that we are convinced of their correctness , and certainly convinced that they warrant further analysis to try to fill the gaps .
finally we note that , for those wishing to focus as rapidly as possible on the calculations for @xmath3 , section 2.3 and the discussion of stirling s theorem in section 3.4 may be omitted en route .
in this paper we use the definition of bernoulli numbers as given recursively by @xmath57 which implies @xmath58 , @xmath59 , @xmath60 , @xmath61 and so on .
for @xmath62 we have @xmath63 and the even - index bernoulli numbers are related to the values of @xmath3 at negative odd integers by @xmath64 .
a generating function for the @xmath65 is given by @xmath66 bernoulli polynomials are then defined by @xmath67 and are monic polynomials satisfying @xmath68 . since @xmath69 we alternatively let @xmath70 be the periodic , period-1 extensions of the bernoulli polynomials from @xmath71 $ ] to all of @xmath72 .
for @xmath73 these are differentiable and satisfy @xmath74 , and this relationship continues to hold for @xmath75 except at the integer points ( of measure @xmath45 ) where @xmath76 is defined to have value @xmath45 .
the first few @xmath77 and @xmath70 are given by @xmath78 and so on where @xmath79 is the fractional part of @xmath80 .
we denote by @xmath81 the polynomials in @xmath82 given by the sums of @xmath83 powers : @xmath84 which turn out to be closely related to the bernoulli polynomials .
we take the fourier transform and inverse fourier transform to be defined by @xmath85(\xi)=\int_{-\infty}^{\infty}f(x)\cdot\textrm{e}^{-ix\xi}\,\textrm{d}x\qquad\textrm{and}\qquad\mathcal{f}^{-1}[g](x)=\frac{1}{2\pi}\int_{-\infty}^{\infty}g(\xi)\cdot\textrm{e}^{ix\xi}\,\textrm{d}\xi\ ] ] finally we denote the odd heaviside function by @xmath86 so that @xmath87 while @xmath88 denotes the standard heaviside function @xmath89
in [ 1 ] the notion of cesaro convergence was recast in terms of a cesaro operator @xmath90(x):=\frac{1}{x}\intop_{0}^{x}f(t)\,\textrm{d}t$ ] ( on functions @xmath91 ) and extended by consideration of the eigenvalues and eigenfunctions of @xmath92 so that a classically divergent function @xmath33 has generalised cesaro limit @xmath93 if @xmath94(x)\rightarrow l$ ] as @xmath95 classically for some regular polynomial in @xmath92 , @xmath96 ( regularity being equivalent to having @xmath97 ) .
since the eigenfunctions of @xmath92 , namely @xmath98 with eigenvalue @xmath99 ( @xmath100 ) , arise naturally in the euler - mclaurin asymptotic expansion for the partial sums of the riemann zeta series , @xmath101 , generalised cesaro summation can be used ( see [ 1 ] ) to derive the analytic continuation of @xmath3 to @xmath102 .
while various `` schemes '' were considered in [ 1 ] , for @xmath3 the cleanest extension arose from a geometric cesaro approach in which a partial sum function ( of a real variable ) is obtained by adding in each term @xmath103at the point @xmath104 .
it is more natural , however , to view this analysis in [ 1 ] as a special case of remainder cesaro summation in which contributions occur at points located relative to an initial point @xmath11 .
specifically , given a function @xmath105 we wish to define its strict remainder sum at @xmath11 by@xmath106(z_{0}):=\sum_{n=1}^{\infty}f(z_{0}+n)\label{eq : remsum}\ ] ] where we want to understand the sum on the rhs in a generalised cesaro sense along the horizontal contour @xmath107 .
adapting the working definition in lemma 4 in [ 1 ] , we do this by adopting the following working definition of cesaro convergence along a contour : [ [ definition-1 ] ] definition 1 : + + + + + + + + + + + + + _ suppose _ @xmath108 _ is a contour parametrised by arc length _
@xmath109 , _ starting at _ @xmath7 _ with _
@xmath110 _ as _
@xmath111 , _ and suppose _
@xmath33 _ is a function on _
@xmath112 _ which can be written as _
@xmath113 _ for some finite collection of constants _
@xmath114 , @xmath115 _ and _ @xmath116 _ and some remainder function _ @xmath117 .
_ if there exists _
such that _
@xmath119(t)\rightarrow l$ ] _ as _
@xmath111 , _ then we say that _ @xmath33 _ has generalised cesaro limit _ @xmath93 _ along _ @xmath112 , _
where _ @xmath92 _ is now the averaging operator _ ( _ in _
@xmath109 ) _ along _ @xmath112 _ defined by
_ @xmath120(t)=\frac{1}{t}\int_{0}^{t}h(u)\,\textrm{d}u\ ] ] [ [ notes - on - definition-1 ] ] notes on definition 1 : + + + + + + + + + + + + + + + + + + + + + + \(i ) in the rest of this paper the only contours we shall need to consider are either horizontal rays parallel to the real axis ( @xmath121 ) or vertical rays parallel to the imaginary axis ( @xmath122 ) , which simplifies interpretation . for these cases
it is not hard to see uniqueness of limits in definition 1 .
\(ii ) it is critical in this definition , and will be crucial throughout this paper , that the functions @xmath123 that we `` throw away '' ( i.e. assign generalised cesaro limit @xmath45 for @xmath124 ) are functions of the _ geometric _ variable @xmath125 , not just the arc - length parameter @xmath109 ; we continue to think of these as cesaro eigenfunctions and generalised eigenfunctions in this setting even though this association is now loose given the necessary definition of @xmath92 in terms of @xmath109 rather than @xmath52 .
of course , in the case where @xmath126 the need for this distinction disappears , at least when @xmath112 is a simple ray as per ( i ) , since in this case a straightforward taylor expansion allows expression of each @xmath123 as a linear combination of functions of the form @xmath127 with all @xmath128 , and vice - versa . when @xmath129 is an integer , however , this distinction becomes pivotal in order to avoid pure powers of @xmath130 , which correspond to cesaro `` eigenfunctions '' with eigenvalue @xmath131 and thus have no generalised cesaro limit ( see [ 1 ] , section 2 ) .
\(iii ) in relation to this last remark , recall the key point that cesaro definition of generalised convergence is fundamentally intended as a tool for constructive analytic continuation ( in @xmath7 and some parameter @xmath29 which drives the values of the @xmath129 s in the situation being analysed ) . as such , the above distinction could be rephrased as saying that when all @xmath126 the definition 1 can be recast , at least for @xmath112 a ray , purely in terms of the parameter @xmath109 ( and then fully justified rigorously in terms of genuine eigenfunctions and eigenvalues of a regular polynomial in @xmath132 ) ; but when @xmath0 is such that there is a @xmath133 we find that we need to distinguish between @xmath52 and @xmath109 as per the discussion in ( ii ) in order to obtain the correct analytic continuation across this value of @xmath0 .
\(iv ) when @xmath112 is just a horizontal ray then , for @xmath7 real , definition 1 devolves simply to the case considered in [ 1 ] and so simply represents a natural extension of the definition in [ 1 ] to achieve the desired constructive analytic continuation from @xmath72 to @xmath134 just mentioned . the case of vertical rays
can be viewed similarly .
while the motivating example of remainder summation in ( [ eq : remsum ] ) leads to a partial sum function with evenly spaced jumps , the function @xmath33 in definition 1 obviously need not have any jumps or even arise from a summation process , or even if it does , may arise in a way where the spacing between summands is not even .
\(v ) we will return briefly to discuss definition 1 and the above observations ( i)-(iv ) further in the final section of the paper , by which stage the intervening calculations will hopefully have both demonstrated the utility of this definition and clarified the meaning and significance of some of these comments ( which may seem obscure at present ) .
however , since our primary focus in this paper is on calculation , we defer further discussion here and simply conclude by summarising definition 1 into the following `` working recipe '' for calculation of generalised cesaro limits along contours , as it will be used throughout the paper to perform the constructive analytic continuation ( in @xmath7 and @xmath0 ) just mentioned : \(a ) first remove linear combinations of cesaro eigenfunctions and generalised eigenfunctions , but critically doing so geometrically as eigenfunctions in @xmath125 , rather than simply @xmath109 , and then \(b ) apply a suitable power of the cesaro averaging operator `` along the contour '' ( i.e. averaging in the contour arc - length @xmath109 ) .
returning now to the case of remainder summation in ( [ eq : remsum ] ) , here we form the contour @xmath107 , form the partial sum function @xmath135 and use definition 1 to obtain @xmath136(z_{0})$ ] as @xmath137 , where @xmath138 and we have used the notation @xmath139 rather than @xmath140 to emphasise that it is eigenfunctions in @xmath52 rather than @xmath109 that we remove from @xmath141 prior to averaging as discussed above . in similar fashion we define @xmath142(z):=\sum_{n=0}^{\infty}f(z+n)$ ] , @xmath143(z):=\sum_{n=1}^{\infty}f(z - n)$ ] , @xmath144(z):=\sum_{n=-\infty}^{\infty}f(z+n)$ ] and so forth . in the last case with a bi - directional sum , the definition is via two independent cesaro sums , for @xmath145 and @xmath146 respectively . in the case of @xmath147
we see at once that @xmath148(0)$](0)$ ] where @xmath147 ; we shall use this notational shorthand extensively ] and the analysis referred to in [ 1 ] consisted of detailing the generalised cesaro interpretation of this remainder summation at @xmath45 for arbitrary @xmath149 .
is playing the role of the variable @xmath0 in the discussion above ; with @xmath150 held fixed the cesaro convergence is used to analytically continue the expression @xmath151(0)$ ] from its classical region of convergence when @xmath152 to all @xmath153 . ] for @xmath154 , @xmath151(z)$ ] in turn yields the hurewicz zeta function @xmath155 via identical cesaro analysis .
as a first simple example of the flexibility allowed by letting @xmath156 and moving to the remainder cesaro viewpoint , consider in this context the polynomials in @xmath157 defined in section 1.1 by @xmath158 .
these are related by formal differentiation of the discrete summation variable , @xmath159 , namely @xmath160 from which they may be readily recursively generated and shown to be connected to the bernoulli polynomials @xmath77 .
this formal differentiation may be rigorised in the remainder cesaro approach by naturally defining @xmath161 , for arbitrary @xmath52 , as @xmath162(0)-r_{+}[\tilde{z}^{-s}](z)=\zeta(s)-\zeta_{h}(z , s)\label{eq : psumzeta}\ ] ] where @xmath163 .
utilising the fact that we may now rigorously differentiate with respect to @xmath52 and that @xmath164 ( after noting that summation and differentiation may be commuted in this case in a careful cesaro analysis ) we immediately obtain the claimed relationship on setting @xmath165 .
a second way in which remainder summation immediately expands the scope of analysis is that @xmath166 ( and @xmath146 etc ) are now operators on functions and thus amenable to spectral analysis .
the notion of a cesaro remainder product follows naturally via exponentiation of the remainder sum of the logarithm:@xmath167(z):=\exp(r_{+}[\ln(f)](z))\label{eq : remprod}\ ] ] for example , for the identity function @xmath10 , we have @xmath168(z)=\exp(r_{+}[\ln(\tilde{z})](z))$ ] and the calculation of @xmath169(z)$ ] by cesaro means follows readily from the euler - mclaurin sum formula .
specifically , for any given @xmath7 , using the notation from [ 1 ] where @xmath170 and recalling that @xmath171 and @xmath52 are generalised eigenfunctions of @xmath92 where @xmath172 , we have@xmath173 thus @xmath174(z_{0})=\exp(c_{z_{0}})\label{eq : remprodza}\ ] ] where@xmath175 it turns out this example leads directly to the @xmath2 function : [ [ lemma-1 ] ] lemma 1 : + + + + + + + + + _ the _ @xmath2 _ _ function is given directly in terms of remainder cesaro products by__@xmath176(0)}{\prod_{r}[\tilde{z}](z)}\label{eq : gammadef}\ ] ] _ _ or equivalently__@xmath177(0)-r_{+}[\ln\tilde{z}](z)=\frac{1}{2}\ln(2\pi)-r_{+}[\ln\tilde{z}](z)\label{eq : lngammadef}\ ] ] this is readily proven by , for example , comparing taylor series for the two sides of ( [ eq : lngammadef ] ) , but we omit details here .
note , however , that it is critical in the above cesaro analysis that summands are introduced geometrically at @xmath178 and cesaro eigenfunctions are then viewed w.r.t .
this geometric variable @xmath179 , rather than simply the summation index @xmath180 ; this need to respect the geometric location of summands in the cesaro setting will be crucial later in this paper .
note that the defining relation for the @xmath2 function , @xmath181 , is self - evident from this definition , as is the fact that @xmath182 for @xmath157 .
other basic identities for @xmath2 ( and for other functions ) likewise follow directly from geometric considerations within the cesaro viewpoint . to conclude section 2 we briefly give three examples to illustrate the way geometry can be exploited in the remainder cesaro framework .
the well - known duplication formulae for @xmath2 state that for any @xmath183@xmath184 using the cesaro definitions in ( [ eq : gammadef ] ) and ( [ eq : lngammadef ] ) this follows almost immediately from the dilation invariance of cesaro summation ( see appendix 5.1 ) . for , taking logarithms , ( [ eq : gammadupl ] ) is equivalent to@xmath185 which , in light of ( [ eq : lngammadef ] ) , is equivalent to@xmath186(z)=(z+\frac{1}{2})\ln n-\sum_{j=1}^{n}r_{+}[\ln\tilde{z}](\frac{z}{n}-\frac{(n - j)}{n})\ ] ] but , by dilation invariance , if we dilate each remainder cesaro sum on the rhs by @xmath187 ( i.e. place each summand @xmath188 at @xmath189 ) and note that @xmath190 , then the summands now intersperse perfectly ( see figure 1 in appendix 5.3 ) to yield@xmath191(z)\ ] ] the result then follows on recalling that @xmath192 is a constant and therefore @xmath193(z)$ ] = ( -z-@xmath194 ( which itself follows immediately on setting @xmath195 in ( [ eq : psumzeta ] ) and using the fact that @xmath196 and @xmath197 ) .
it is clear from the structure of the above proof that an analogous duplication identity will hold for any function defined by remainder summation @xmath136(z)$ ] ( or corresponding product ) as long as @xmath198 can be related readily to @xmath199 .
it is trivial for example that@xmath200 here the roots of @xmath33 are all simple roots of multiplicity 1 at the points @xmath293 , @xmath82 , and so for @xmath17 at arbitrary @xmath7 in ( [ eq : genrootidv1 ] ) , we have@xmath294 these are equal by a well - known identity , and so @xmath19 satisfies the generalised root identities ( [ eq : genrootidv1 ] ) for arbitrary @xmath7 and all @xmath12 .
since the roots of @xmath33 are closely related to the positive integers , the rhs of ( [ eq : genrootidv1 ] ) leads easily to values of @xmath295 , @xmath118 .
in particular , for @xmath296 the root identity at @xmath150 immediately captures euler s famous formula that @xmath297 .
this is an exampe of an `` inverse '' use of the generalised root identities , where a function of interest , @xmath3 , arises naturally from the root side of the identities for a known function , @xmath19 , and information about @xmath3 is then deduced by considering the derivative side of the identities for @xmath33 .
we shall return to this example , along these lines , in the next section .
the functional equation for @xmath2 states that@xmath201 or equivalently that@xmath202 this can be proven in the remainder cesaro framework using ( [ eq : lngammadef ] ) by converting the pair of remainder sums on the lhs in ( [ eq : lngammafunceqn ] ) into a bi - directional sum , @xmath203 , and then noting that this is automatically periodic with period 1 and thus amenable to fourier analysis . in more detail , rewriting ( [ eq : lngammadef ] ) as@xmath204(z),\ ] ] consider the diagram in figure 2 in appendix 5.3 , for a given choice of @xmath7 .
now in the case shown , for @xmath205 , and using @xmath206 as the principal branch of @xmath130 , we have @xmath207 for all @xmath180 .
let @xmath208 be the variable in the partial sum , @xmath209 , for the cesaro computation of @xmath210(1-z_{0})$ ] and let @xmath211 be the corresponding variable in the partial sum , @xmath212 , for computing @xmath213(z_{0})$ ] , so that we have@xmath214(1-z_{0})=\underset{z_{1}\rightarrow\infty}{clim}\ , s_{1}(z_{1})\quad\textrm{and}\quad r_{-}[\ln\tilde{z}](z_{0})=\underset{z_{2}\rightarrow-\infty}{clim}\ , s_{2}(z_{2})\ ] ] then@xmath215 now we know from ( [ eq : remprodzb ] ) that@xmath216 therefore@xmath217+(z_{0}-k)+c_{-z_{0}}+i\pi k+o(1)\ ] ] but@xmath218 thus@xmath219 and on adding @xmath209 and @xmath212 we have@xmath220(z_{0})=i\pi(z_{0}-\frac{1}{2})+\ln(2\pi)-\ln(\gamma(1-z_{0}))-\ln(\gamma(z_{0}))\label{eq : lngammabidirectional}\ ] ] we see that ( [ eq : lngammafunceqn ] ) follows if and only if@xmath220(z)=\ln(\sin(\pi z))+i\pi(z-\frac{1}{2})+\ln2\ ] ] and on writing@xmath221 this in turn follows if and only if@xmath220(z)=-\sum_{n=1}^{\infty}\frac{e^{2\pi inz}}{n}\label{eq : lngammabidirectional2}\ ] ] now , by definition , @xmath222(z)$ ] is a periodic function with period 1 , expressible as a fourier series@xmath220(z)=\sum_{n=-\infty}^{\infty}a_{n}e^{2\pi inz}\ ] ] and so we see that ( [ eq : lngammabidirectional2 ] ) is equivalent to verifying that the fourier coefficients in this expansion , @xmath223 , satisfy@xmath224 but@xmath225(z)\cdot e^{-2\pi inz}\,\textrm{d}z\\ & = & \intop_{-\infty}^{\infty}\ln z\cdot e^{-2\pi inz}\,\textrm{d}z\\ & = & \intop_{-\infty}^{\infty}\ln|z|\cdot
e^{-2\pi inz}\,\textrm{d}z+i\pi\intop_{-\infty}^{0}e^{-2\pi inz}\,\textrm{d}z\end{aligned}\ ] ] in a cesaro analysis the second term here gives@xmath226 while the first term , on performing integration by parts and using a cesaro calculation to justify ignoring the boundary terms , gives@xmath227 now the fourier transform of the constant function 1 is @xmath228 and multiplication by @xmath52 corresponds to @xmath229 , so division by @xmath52 corresponds to @xmath230 and we see that this last term integrates to a heaviside function@xmath231 combining terms , this yields at once the required form for @xmath223 and the functional equation for @xmath2 follows as claimed from ( [ eq : lngammabidirectional2 ] ) .
we see that the relationship between values of @xmath2 at @xmath52 and @xmath232 expressed in the functional equation ( [ eq : gammafunceqn ] ) arises naturally from the geometry of the remainder cesaro approach shown in figure 2 , together with the fourier - amenability of the resulting periodic bi - directional sum @xmath203 . as such it is again
the case that an analogous functional equation should hold for any function defined , like @xmath2 , by a remainder cesaro sum .
this is indeed the case , for example , for the hurewicz zeta function @xmath233 .
while the fourier leg of the analysis is difficult to perform in this case for arbitrary @xmath52 , for the particular case of @xmath234 it can be accomplished relatively directly and the functional equation for the riemann zeta function is then derived , namely@xmath235 for reasons of space , however , we omit details here , particularly in light of the fact that we will derive the functional equation for @xmath3 in section 3.3 by alternative means arising from our introduction of generalised root identities . note , however , that in the derivation using bi - directional sums and fourier theory the connection between values at @xmath149 and @xmath236 arises from the fourier theory rather than from the cesaro geometry as it did in the example of @xmath2
. the roots of @xmath21 are all simple poles ( @xmath298 ) at @xmath299 .
thus for @xmath17 and arbitrary @xmath7 in ( [ eq : genrootidv1 ] ) , we have@xmath300 this is divergent for all @xmath7 and is not even cesaro summable because the partial sums involve the generalised cesaro eigenfunction with eigenvalue 1 , @xmath301 .
however , since this cesaro - obstruction is uniform , it can be `` renormalised '' away uniformly as@xmath302 where @xmath303 is euler s constant and arises because the difference between the @xmath301 divergence we are removing and the partial sums of @xmath304 approaches @xmath112 in the limit ( @xmath305 ) .
but ( [ eq : renormgammamu1 ] ) is in turn a well - known expression for @xmath306 and thus we see that , after renormalisation of the ln cesaro divergence , @xmath21 does satisfy the generalised root identity ( [ eq : genrootidv1 ] ) for @xmath17 and arbitrary @xmath7 ( note that the expression in ( [ eq : renormgammamu1 ] ) retains the required simple poles at @xmath307 ) .
the need to perform this renormalisation adjustment arises only for the case of @xmath17 . for @xmath296 the rhs of ( [ eq : genrootidv1 ] )
is @xmath308 , which is classically convergent for all @xmath7 and clearly equals the the lhs , namely @xmath309 , on differentiating the expression for @xmath306 in ( [ eq : renormgammamu1 ] ) ; likewise for @xmath310 .
thus overall @xmath21 satisfies the generalised root identities ( [ eq : genrootidv1 ] ) for arbitrary @xmath7 and all @xmath12 , albeit after requiring an additional renormalisation when @xmath17 to uniformly remove the cesaro non - amenable ln - divergences in this case .
it is easy to see that the same continues to hold true for the case of general @xmath311 . for @xmath21 , on the root side of ( [ eq : genrootidv2 ] )
we clearly have at once@xmath402 when @xmath150 this reduces to the zeta function@xmath403 and since @xmath404 , so for example at @xmath405 we get @xmath406 and in general for @xmath118 @xmath407 on the derivative side , noting @xmath408 and recalling the identity in ( [ eq : renormgammamu1 ] ) we have @xmath409 for all except @xmath17 we can omit the @xmath392 terms arising from the @xmath410 and @xmath112 terms here and thus , by the fourier identities canvassed in ( v ) , we obtain @xmath411 on ignoring the contributions from the @xmath325 term in @xmath412 as just discussed . thus @xmath413+\\ \left[\frac{(i\xi)}{2!}+\frac{(i\xi)^{2}}{3!}+\ldots\right]^{2}-\ldots\end{array}\right\ } \textrm{d}\xi\\ \\ & = & \frac{i}{\gamma(\mu)}\int_{0}^{\infty}(i\xi)^{\mu-2}\left\ { \begin{array}{cc } 1+i\left[(z_{0}+\frac{1}{2})\right]\xi+\left[-\frac{1}{2}z_{0}^{2}-\frac{1}{2}z_{0}-\frac{1}{12}\right]\xi^{2}\\ \\+(-i)\left[\frac{1}{6}z_{0}^{3}+\frac{1}{4}z_{0}^{2}+\frac{1}{12}z_{0}\right]\xi^{3}\\ \\+\left[\frac{1}{24}z_{0}^{4}+\frac{1}{12}z_{0}^{3}+\frac{1}{24}z_{0}^{2}+\frac{1}{120}\right]\xi^{4}+\ldots\end{array}\right\ } \textrm{d}\xi\end{aligned}\ ] ] now suppose @xmath1 . working distributionally as discussed earlier in ( iii ) , since for @xmath1 we have @xmath414 ( by ( [ eq : hormanderlemma ] ) ) , so for @xmath42 we have immediately @xmath415 , in agreement with the first result from ( [ eq : gammagenmuc ] ) . similarly , for @xmath416 and @xmath417 , after disentangling the impact of the @xmath418 factor and the factorial arising from differentiation , we get agreement with the formulae in ( [ eq : gammagenmuc ] ) ( each arising as a scalar multiple of the appropriate coefficient of @xmath419 in the bracket ) ; and in general @xmath420 so that @xmath21 does satisfy the generalised root identities ( [ eq : genrootidv2 ] ) for all @xmath1 .
thus , overall , @xmath21 satisfies the generalised root identities ( [ eq : genrootidv2 ] ) for arbitrary @xmath7 and @xmath421@xmath422 ; and in fact , as noted for @xmath264 in a previous section , it is easy to adapt the above arguments to verify that this remains true for @xmath311 for any @xmath423@xmath134 . to conclude our analysis of @xmath21 we illustrate the power of these new root identities for @xmath1 to give information about asymptotic behaviour
we do this by constructively deducing stirling s theorem from successive consideration of the identities for @xmath424 - 1,-2,@xmath425 , namely : [ [ stirlings - theorem ] ] stirling s theorem : + + + + + + + + + + + + + + + + + + + _ as _ @xmath39 _ we have _
@xmath426 _ where _ @xmath427 _ has the asymptotic expansion _
@xmath428 _ equivalently , as _
@xmath39 , _ we have asymptotically _
@xmath429 to deduce this consider first the generalised root identity for @xmath42 . by ( [ eq : gammagenmuc ] ) we have seen ( from a cesaro count of roots on the rhs ) that this means@xmath430 as we move through @xmath431 the corresponding polynomials in ( [ eq : gammagenmuc ] ) are successive integrals of this ( as one would expect from each additional application of @xmath432 , i.e. integration , on the lhs ) , with the only new information each time being the value of the integration constant , namely @xmath433 ( by ( [ eq : gammagenmud ] ) ) .
our requirement is thus to understand ( a ) what factors must occur in @xmath434 in order for the terms @xmath7 and @xmath325 to arise on the lhs in ( [ eq : stirlingproofa ] ) when @xmath42 , and then ( b ) how the successive integration constants just identified can be made to appear on the derivative side of the root identities ( [ eq : genrootidv2 ] ) when @xmath431 by the inclusion of further terms without disturbing the identities already examined .
now considering the definition in ( [ eq : complexderivdef ] ) we see that integer powers of @xmath7 will arise when the action of @xmath435(\xi)\bullet d\xi$ ] on @xmath436 consists of differentiation to some integer order . in light of ( [ eq : hormanderlemma ] ) this means we require @xmath437(\xi)=\chi_{+}^{-n}(\xi)=\frac{1}{\gamma(1-n)}\frac{h_{0}(\xi)}{\xi_{+}^{n}}$ ] for some @xmath118 .
we thus require @xmath438(\xi)$ ] itself to be of the form @xmath439 for some suitable @xmath440 and @xmath441 ( ignoring the contribution from the @xmath325 in @xmath412 as usual ) .
in particular it is clear that when @xmath424 in order to get a constant term @xmath325 we need to have @xmath438(\xi)=\frac{1}{2}\frac{\tilde{h}_{0}(\xi)}{\xi}$ ] ( so that we get @xmath442 , i.e. @xmath443 , acting on @xmath436 ) ; while to get a term @xmath7 we need to have @xmath438(\xi)=\frac{\tilde{h}_{0}(\xi)}{\xi^{2}}$ ] ( so that we get @xmath444 , i.e. @xmath445 , acting on @xmath436 ) .
but , comparing with ( [ eq : fourierresults1 ] ) , it is clear that this means we need to take @xmath446 and so @xmath447 must appear in @xmath434 .
these in turn then generate the terms @xmath448 in the @xmath43 identity in ( [ eq : gammagenmuc ] ) , leaving only the @xmath449 term to account for . for this , in the same way we used terms @xmath450 to `` heal '' obstructions for root identities with @xmath264 , so we now naturally consider terms of the form @xmath451 to `` heal '' corresponding obstructions when @xmath1 .
taking @xmath452 first - on taking logs this becomes @xmath453 and , by ( [ eq : fourierresults1 ] ) , @xmath438(\xi)=-2\pi ia\tilde{h}_{0}(\xi)$ ] .
thus , when @xmath43 , @xmath454(\xi)=-\frac{2\pi a}{\xi_{+}}$ ] ( eliding @xmath455 and @xmath456 as usual ) and on writing @xmath457 we see from ( [ eq : hormanderlemma ] ) that we will get a contribution of @xmath458 from this term in the root identity ( arising from @xmath459 acting on @xmath436 ) . in order to match the required @xmath449 we thus take @xmath460 and
so include an extra factor of @xmath461 in @xmath21 ( or @xmath462 in @xmath434 ) in order to satisfy the @xmath43 root identity .
importantly , note that this extra term does not disturb the existing identity at @xmath42 since there it leads to a finite cesaro integral which is cancelled to @xmath45 by the @xmath463 factor in the @xmath42 identity .
so far we thus have contributions @xmath464 in @xmath465 thereby ensuring the @xmath42 and @xmath43 identities are satisfied . in the same way , working inductively , for @xmath466 the existing terms generated up to @xmath467 will integrate to yield the polynomial @xmath468 in the expression ( [ eq : gammagenmud ] ) for the @xmath466 root identity ( after adjusting for the change in @xmath366 factor on the derivative side ) and we need to introduce a further term @xmath451 in order to match the new constant term @xmath469 which is all that remains unaccounted for .
but then taking logs yields @xmath470 with @xmath438(\xi)=2\pi a\textrm{e}^{-i\frac{\pi}{2}n}\frac{\tilde{h}_{0}(\xi)\xi^{n-1}}{\gamma(n)}$ ] and so @xmath471(\xi)=-a\cdot\frac{n}{\gamma(0)}\frac{1}{\xi_{+}}=-an\delta_{0}(\xi)$ ] after simplifying the two @xmath2 terms by cancellation ( or alternatively using the functional identity ( [ eq : gammafunceqn ] ) ) and retaining only the @xmath456 term from @xmath455 as usual . acting on @xmath436
we thus need to take @xmath472 in order to match the required constant term . when @xmath187 is even ( i.e. @xmath473 )
we are at a trivial zero of @xmath3 and @xmath474 so that there are no even order reciprocal terms in the asymptotic expansion for @xmath427 ; when @xmath475 is odd we know that @xmath476 so that @xmath477 and we have to include a term @xmath478 in our expansion for @xmath21 . and , again , each new term included does not disturb our existing identities for @xmath479 since there the @xmath451 term leads to finite cesaro integrals on the derivative side which cancel to @xmath45 against the @xmath480 factor in the @xmath481 identity , @xmath482 .
consideration of all the root identities for @xmath1 thus leads us constructively to deduce@xmath483 exactly as in ( [ eq : stirlinga ] ) and ( [ eq : stirlingb ] ) , and it remains only to show that @xmath484 .
but , letting @xmath485 through the positive integers , taking logs and noting that @xmath486 as @xmath487 we get@xmath488\right\ } = \lim_{k\rightarrow\infty}\left(\sum_{j=1}^{k}\ln j-\left[(k+\frac{1}{2})\ln k - k\right]\right)\ ] ] but this last expression was evaluated in [ 1 ] ( in the course of calculating @xmath489 ) as precisely @xmath490 and so @xmath484 as required .
this completes the constructive derivation of the expression in ( [ eq : stirlinga ] ) purely from the root identities for @xmath1 satisfied by @xmath21 . to complete the proof of stirling s theorem it remains only to verify that no further terms of the form @xmath450 , @xmath491 , are required in order to also satisfy the root identities for @xmath12 , i.e. that the expression @xmath492 already satisfies all these root identities for @xmath21 without requiring the healing of any obstructions . for @xmath17
we immediately encounter a problem from the term @xmath493 ; on the derivative side of the root identity , after the usual passage from @xmath494 to @xmath495 , this yields an integrand of @xmath496 and this leads to a cesaro divergent integral ( because log - divergent ) in its action on @xmath436 since we no longer have the coefficient @xmath463 to normalise this into a @xmath397-function .
however , this merely mirrors the corresponding cesaro divergence of @xmath497 on the root side , for which we needed to perform the renormalisation described earlier in verifying the @xmath17 root identity ; a corresponding renormalisation is of course required on the derivative side . with this in mind
we proceed with the other terms : @xmath498 gives integrand contribution @xmath499and each term @xmath500 gives integrand contribution @xmath501 ( by ( [ eq : fourierresults1 ] ) ) .
thus , for @xmath502 on the derivative side of the root identity we get @xmath503 recalling the generating function for the bernoulli numbers in ( [ eq : bernoullinumsgenfn ] ) we see that this becomes@xmath504 where the last step follows from a direct cesaro computation of the integrals . since this formally equals the root side of the @xmath17 root identity we see that , ( up to renormalisation on both sides ) , @xmath492 does indeed satisfy the root identity for @xmath21 without requiring any further correction terms of the form @xmath505 . for @xmath296 and higher ,
no renormalisation issues arise . at @xmath296 , on the derivative side , @xmath493 contributes @xmath506 , @xmath498 contributes @xmath507 , @xmath508 makes no contribution and @xmath500 contributes @xmath509 after @xmath510-fold cesaro integration by parts .
we thus end up with @xmath511 which is a well - known expression for @xmath512 .
since we have already verified that this equals the root side of the generalised root identity for @xmath21 ( i.e. that @xmath21 does satisfy the @xmath296 root identity ) so it follows again that @xmath492 satisfies the @xmath296 root identity without need of any further correction factor @xmath513 . since this is true for arbitrary @xmath7 and without need of renormalisation , so it follows also for all @xmath514 and thus stirling s theorem
is finally proven .
one comment is worth noting regarding this constructive argument .
stirling s theorem actually expresses the asymptotic behaviour of @xmath21 only as @xmath39 .
the argument above , however , is naturally `` two - sided '' and suggests that @xmath21 is asymptotically given by @xmath515 also as @xmath38 , since in picking out the form of functions in @xmath341 so as to match the root side of our identities on the derivative side , we have had to call on functions ( like @xmath516 etc ) which are defined on all of @xmath72 , not just @xmath517 or @xmath518 . to check this
, we can use the logarithmic version of the functional equation for @xmath2 in ( [ eq : lngammafunceqn ] ) to deduce the asymptotic behaviour of @xmath434 as @xmath38 from its known behaviour as @xmath381 from stirling s theorem .
we have , as @xmath38 , @xmath519 and , by stirling s theorem , @xmath520\\ \\-\frac{1}{360}\frac{1}{z^{3}}\left[1-\frac{1}{z}+\frac{1}{z^{2}}-\ldots\right]^{3}-\ldots\end{array}\right\ } \\ \\ & = & -(z+\frac{1}{2})\cdot i\pi+\left\ { \begin{array}{cc } ( z+\frac{1}{2})\ln z - z+\frac{1}{12}\frac{1}{z}\\ \\-\frac{1}{360}\frac{1}{z^{3}}-\ldots\end{array}\right\ } \end{aligned}\ ] ] after simplification .
thus , overall , as @xmath38 , we have @xmath521 we see that , as predicted , we do have the same asymptotic behaviour as per stirling s theorem also when @xmath38 , except that in addition we have the linear and constant terms shown ( @xmath522 and @xmath523 respectively ) and the term @xmath524 which contains a countable collection of ln - divergences at the negative integer points . the reason our above derivation of stirling s theorem nonetheless worked in spite of these extra one - sided terms as @xmath38 is that , owing to the @xmath366 factors on the derivative side of our root identities , the contributions made by these terms on this derivative side for @xmath1 are @xmath45 .
we shall omit calculations demonstrating this here , but note that in general terms this reflects the fact that it is asymptotic behaviour within a log - scale , rather than local divergences ( even a countable collection of them ) , that matter in the computation of the derivative side of the root identities for @xmath1 .
in [ 1 ] it was shown that if @xmath237 then we have@xmath238 and that this combined with the euler - mclaurin sum formula leads to the derivation of the values of @xmath3 as cesaro limits of partial sums @xmath239 .
since @xmath240 , so when @xmath241 we obtain@xmath242 and in light of ( [ eq : psumzeta ] ) this can be re - expressed as saying that@xmath243 in this last example we show how ( [ eq : zetahintegral1 ] ) in fact follows for arbitrary @xmath244 as a consequence again of geometric considerations in the remainder cesaro approach together with dilation - invariance arguments similar to those deployed in example 1 for the @xmath2-function duplication formulae .
[ [ lemma-2 ] ] lemma 2 : + + + + + + + + _ _ we have__@xmath245 [ [ proof ] ] proof : + + + + + + suppose initially that @xmath246 . using right riemann sums , the integral on the lhs becomes@xmath247(-\frac{j}{n})\ ] ] but dilating by n and noting that @xmath248 and that the summation points after dilation all interleave to precisely fill up the positive integers , it follows from cesaro dilation - invariance that this becomes@xmath249(0)=\underset{n\rightarrow\infty}{\lim}n^{s-1}\zeta(s)=0\ ] ] this proves ( [ eq : zetahintegral2 ] ) for @xmath246 . at @xmath250 the integral is not cesaro convergent but for all other @xmath149 the result then follows by analytic continuation . :
for example , for any prime @xmath41 , @xmath251 where the sum is over @xmath252 ; for @xmath253 prime , @xmath254 where the sum is over @xmath255 , j coprime to @xmath256 , and so on . ] having introduced and illustrated applications of remainder cesaro summation , we now turn to the primary focus of this paper , in which it plays a key role , namely generalised root identities .
the root set of @xmath36 consists of the trivial zeros at @xmath312 ( with @xmath313 ) , the simple pole at @xmath250 ( with @xmath314 ) and the famous non - trivial zeros in the critical strip @xmath315 ( for which @xmath316 are unknown in general ) .
we let @xmath46 denote the set of trivial roots and @xmath317 the set of non - trivial roots , which we will also denote by @xmath318 rather than @xmath268 .
if the riemann hypothesis is true , then the @xmath318 occur in conjugate pairs solely on the critical line @xmath319 ; if not then some occur in quadruples via reflection in the real axis and critical line : @xmath318 , @xmath320 , @xmath321 , and @xmath322 . of course
, the roots in @xmath317 are not known exactly and thus tackling the root side of the generalised root identities ( [ eq : genrootidv1 ] ) directly is difficult . however , working empirically first , we can take a list of , say , the first 100,000 non - trivial zeros and use them to test experimentally whether @xmath3 seems to satisfy ( [ eq : genrootidv1 ] ) for @xmath17 . in doing this ,
however , a renormalisation analogous to the last example will have to be carried out to handle the trivial zeros , @xmath46 ; considering the case @xmath323 initially , this is equivalent to setting @xmath324 ( the @xmath325 factor arising naturally since @xmath46 only covers the negative even integers ) .
thus , taking initially only the truncated non - trivial root - set @xmath326 given , for example , by odlyzko in [ 2 ] and consisting of the first 100,000 non - trivial zeros ( actually 200,000 including conjugates ) we find that the root side of ( [ eq : genrootidv1 ] ) for @xmath17 at @xmath323 becomes@xmath327 ( on performing the last computation numerically to obtain @xmath328 ) .
by contrast the derivative side in ( [ eq : genrootidv1 ] ) gives @xmath329 .
at once we see that @xmath3 does not directly satisfy the generalised root identities for @xmath17 at @xmath323 , with the obstruction in this instance being , numerically , @xmath330 .
next consider @xmath331 . here , by the symmetry outlined above , @xmath332 and so @xmath333 . in this case
the calculation of the renormalised value of the first sum is best done by noting that the trivial zeros occur at the same locations as the poles of @xmath334 and thus , up to an overall factor of -1 , this sum can be evaluated using the generalised root identity for @xmath334 at @xmath331 , namely as @xmath335 .
we thus obtain @xmath336 while @xmath337 , so that again we obtain the same value of @xmath338 for the obstruction .
this suggests , along the lines discussed before , that while @xmath3 does not directly satisfy the generalised root identities ( [ eq : genrootidv1 ] ) for @xmath17 , the function @xmath25 may well do so .
this conjecture can be expressed in a different way .
recall that the functional equation for @xmath3 given in ( [ eq : zetafunceqn ] ) can be re - expressed as simply@xmath339 since the generalised root identities ( [ eq : genrootidv1 ] ) hold for @xmath340 and @xmath334 ( as discussed in example 2 ) , and since a product of functions satisfying these identities will also satisfy these identities , so the conjecture that @xmath25 satisfies them for @xmath17 and arbitrary @xmath24 is equivalent to having @xmath341 satisfy ( [ eq : genrootidv1 ] ) for @xmath17 and arbitrary @xmath24 .
in fact this conjecture turns out to be true . in light of the discussion of @xmath2 in the previous example it is equivalent to the following corollary which can be found in [ 4 , pg35 ] : [ [ theorem-1 ] ] theorem 1 : + + + + + + + + + + _ let _
@xmath317 _ denote the set of non - trivial zeros of _ @xmath3 , _ and _ @xmath342 _ denote the subset with imaginary part _ @xmath343 _ then , with sums understood to include multiplicities , we have _
@xmath344 thus , overall , both @xmath25 and @xmath26 satisfy the generalised root identities ( [ eq : genrootidv1 ] ) for @xmath17 and arbitrary @xmath24 , and hence also for arbitrary @xmath345 . since the factor @xmath28 only contributes to the root identities when @xmath17 , note that for @xmath345 the root identities will in fact be satisfied directly by @xmath3 ( which can of course be readily spot - checked experimentally ) . note that , as discussed in [ 4 ] , this corollary is in fact the expression of the hadamard product formula for @xmath341 ( which is a holomorphic integral function of order @xmath131 ) .
we thus see that the generalised root identity ( [ eq : genrootidv1 ] ) for @xmath17 is in this case equivalent to the hadamard product formula for @xmath341 .
since the identities for @xmath345 are just derivatives of the @xmath17 identity , they will contain no additional information , but we turn now to considering cases of more general @xmath0 , in particular @xmath1 , in the hope that they will lead to new tools with additional content beyond the hadamard identity . in ( [ eq : genrootidv1 ] ) our current root identities
constitute a generalisation from the case of polynomials , relating derivatives of @xmath346 at @xmath7 to sums of integer powers of shifted reciprocals of ( generalised ) roots of @xmath33 .
it is natural , however , to ask whether they can be further generalised by allowing @xmath0 to be an arbitrary complex number rather than just a positive integer ; that is@xmath347 the lhs here we call the derivative side of the root identity and denote by @xmath348 ( or just @xmath349 if the context is clear ) .
the rhs is the root side and is denoted by @xmath350 ( or just @xmath351 ) , and ( [ eq : genrootidv2 ] ) is interpreted as asserting identity of these two functions of two complex variables .
of course , in attempting such an extension the question immediately arises of how to interpret either side of ( [ eq : genrootidv2 ] ) : ( a)(i ) to overcome the potential divergence of the sum on the root side ( e.g. when @xmath352 becomes negative ) the rhs must be interpreted via a generalised convergence scheme , which analytically continues the rhs from its region of convergence in the @xmath0-plane . in all cases in this paper
this will be a generalised cesaro scheme .
( a)(ii ) in implementing this cesaro approach , the sum on the root - side will need to be interpreted geometrically .
specifically , each term @xmath30 must be added in at the shifted point @xmath31 itself in the complex plane ( rather than , for example , always being added in at @xmath268 itself ) .
thus , as @xmath7 varies , not only do the summands vary , but their locations move too
. we shall emphasise this by writing @xmath353 on the rhs in ( [ eq : genrootidv2 ] ) .
( b)(i ) on the derivative side in ( [ eq : genrootidv2 ] ) , the interpretation of @xmath32 also becomes problematic .
two facts that should hold in any such definition , however , are that@xmath354 and@xmath355 we shall formalise the definition of the lhs of ( [ eq : genrootidv2 ] ) in a manner consistent with these criteria below .
first , however , we demonstrate the extra power obtained from this extension of the generalised root identities to arbitrary complex @xmath0 by showing how the full functional equation for @xmath3 , given in ( [ eq : zetafunceqn ] ) , follows very simply from applying them to the function @xmath19 considered before in example 1 in section 2.1 . for in this case , taking @xmath150 in ( [ eq : genrootidv2 ] ) we have@xmath356
while for @xmath357 we have@xmath358 so that setting @xmath359 we obtain@xmath360 which is precisely ( [ eq : zetafunceqn ] ) . thus the functional equation for @xmath3 is equivalent to @xmath292 satisfying the generalised root identities ( [ eq : genrootidv2 ] ) ( after also extending easily to the case of @xmath17 ) . with this example as a motivation , and ( [ eq : muderiva ] ) and ( [ eq : muderivb ] ) as guidance , we formalise the meaning of the complex derivative @xmath32 in the lhs of ( [ eq : genrootidv2 ] ) via fourier theory , in a manner familiar from linear partial differential operators , as follows : ( b)(ii)writing @xmath33 as the inverse fourier transform of its fourier transform , namely@xmath361 we define @xmath362 by@xmath363(\xi)\textrm{e}^{iz_{0}\xi}\textrm{d}\xi\label{eq : complexderivdef}\end{aligned}\ ] ] with ( a)(i ) , ( a)(ii ) and ( b)(ii ) the generalised root identities ( [ eq : genrootidv2 ] ) now have a well - defined meaning for arbitrary complex @xmath0 .
we have considered their interpretation for @xmath19 and will examine them next for @xmath21 and @xmath36 , but first we make several general observations about these identities and their potential uses : \(i ) while the cases of @xmath264 give information on the sums of integer powers of shifted reciprocals of roots , ( [ eq : genrootidv2 ] ) gives a much more sensitive relationship between the distribution of these roots and the behaviour of the derivatives of the log of the function .
\(ii ) in paricular , for @xmath42 , ( [ eq : genrootidv2 ] ) should give information about the cesaro count of the roots ( adding @xmath267 at each point @xmath364 ) , while for @xmath365 we get information regarding the cesaro sums of first , second and higher powers of these roots . in particular , taking @xmath352 sufficiently negative should give information about the asymptotic distribution of the roots and the cases @xmath1 will be of particular interest .
\(iii ) for @xmath1 , however , it is immediately clear that further care will need to be taken in ( [ eq : genrootidv2 ] ) owing to the poles of @xmath2 in the factor @xmath366 on the lhs . at first glance
these appear to make the @xmath348 always zero whenever@xmath1 which seems problematic ( e.g. in interpreting the case @xmath42 as yielding a count of roots ) .
thus , in general , we will need to interpret ( [ eq : genrootidv2 ] ) distributionally , and for @xmath1 the distributional result we shall rely on in this paper is the following fact ( see hrmander s treatment in [ 3 , pp74 ] ) regarding the function@xmath367 and its normalised counterpart @xmath368 : [ [ lemma-3 ] ] lemma 3 : + + + + + + + + _ for _ @xmath369 _ _ we have__@xmath370 _ where _ @xmath371 _ is the usual delta distribution_. \(iv ) the fact that ( [ eq : genrootidv2 ] ) naturally gives ( cesaro ) asymptotic information about the distribution of the roots of @xmath33 makes it a natural tool with which to investigate this asymptotic behaviour . another way of looking at this is to compare with the case of ( [ eq : genrootidv1 ] ) in which @xmath0 was restricted to @xmath372 . from an experimental point of view , since for any positive integer @xmath0 the rhs of ( [ eq : genrootidv1 ] ) or ( [ eq : genrootidv2 ] ) diverges as @xmath7 approaches any finite root @xmath268 , so the cases of such positive integer @xmath0 give a tool for detecting the location of finite roots by looking at the dominant behaviour of these identities as we let @xmath7 approach such a location . they give little immediate guidance , however , about how the root locations behave asymptotically as they tend to @xmath373 .
by contrast , for @xmath42 the rhs in ( [ eq : genrootidv2 ] ) is clearly insensitive to the location of finite roots , but its cesaro calculation depends critically on the asymptotic location of the roots as they approach @xmath373 ; and the same is qualitatively true for @xmath374 . in part
this in turn reflects the non - locality of the derivatives @xmath32on the lhs whenever @xmath375 , which can briefly be seen by considering an alternative approach to defining such derivatives , namely : let @xmath376 be the operator of translation by @xmath277 ( i.e. @xmath377(z)=g(z+h)$ ] ) .
since @xmath378 , so in general formally@xmath379 it is only for @xmath12 that this series definition truncates and leaves a purely local resulting function ; for all other @xmath0 the resulting function , @xmath380 , depends on @xmath291 as @xmath381 and so is intrinsically non - local .
we get @xmath382which in the limit yields the left riemann sum definition of @xmath383 . ]
\(v ) in using the fourier definition of the lhs in ( [ eq : genrootidv2 ] ) we shall use the standard fourier relationships repeatedly , namely @xmath384(\xi)=i\mathcal{f}[f]^{\prime}(\xi)\quad\;\&\;\quad\mathcal{f}[f^{\prime}(x)](\xi)=i\xi\mathcal{f}[f](\xi)\ ] ] @xmath385(\xi)=f[f](\xi - a)\quad\;\&\;\quad\mathcal{f}[f(x+a)](\xi)=\mathcal{f}[f](\xi)\textrm{e}^{ia\xi}\ ] ] since @xmath386=2\pi\delta_{0}(\xi)$ ] it follows that , respecting oddness / evenness , we have @xmath387=-2\pi i\tilde{h}_{0}(\xi)=-2\pi i\cdot(h_{0}^{+}(\xi)-\frac{1}{2})\ ] ] and @xmath388=-2\pi\tilde{h}_{0}(\xi)\cdot\xi=-2\pi(h_{0}^{+}(\xi)-\frac{1}{2})\cdot\xi\ ] ] and in general@xmath389}=2\pi e^{-\frac{i\pi\rho}{2}}\,\frac{\tilde{h}_{0}(\xi)\xi^{\rho-1}}{\gamma(\rho)}=2\pi e^{-\frac{i\pi\rho}{2}}\,\frac{(h_{0}^{+}(\xi)-\frac{1}{2})\xi^{\rho-1}}{\gamma(\rho)}\label{eq : fourierresults1}\ ] ] also , inverting the relationship for fourier transform of a derivative and applying it to the formula for @xmath390 $ ] above , we have @xmath391=-2\pi\frac{\tilde{h}_{0}(\xi)}{\xi}=-2\pi\frac{(h_{0}^{+}(\xi)-\frac{1}{2})}{\xi}\label{eq : fourierresults2}\ ] ] and=-2\pi\frac{\tilde{h}_{0}(\xi)}{\xi}-2\pi\gamma\delta_{0}(\xi)$ ] with the extra term arising from the constant of integration ; however we ignore the additional @xmath392 term here since it will not contribute to the derivative side of the root identities at @xmath1 in any of our remaining calculations in this paper on account of the factor @xmath366 whose denominator diverges at such @xmath0 . ]
@xmath393=2\pi i\,\frac{\tilde{h}_{0}(\xi)}{\xi^{2}}=2\pi i\,\frac{(h_{0}^{+}(\xi)-\frac{1}{2})}{\xi^{2}}\quad\;\&\ldots\label{eq : fourierresults2a}\ ] ] with corresponding results for higher anti - derivatives on the lhs .
it is these latter identities that will be used most directly in what follows . from the point of view of calculation on the derivative side of the root identities ( [ eq : genrootidv2 ] ) , note that @xmath394 so that , for @xmath395 , the first term in the brackets in these expressions leads naturally to use of the hrmander result ( [ eq : hormanderlemma ] ) .
the second half of these expressions , @xmath396 , leads , under inverse fourier transform , either to @xmath397-function type contributions ( for @xmath398 ) which may be ignored for @xmath399 or to heaviside - type contributions in @xmath7 ( for @xmath395 ) whose finite values for any @xmath7 cancel to @xmath45 under the factor @xmath366 on the derivative side when @xmath1 .
as such , for our calculations when @xmath1 in the root identities for @xmath2 in the next section we may effectively ignore these second - half contributions and elide the distinction between @xmath400 and @xmath401 .
we now take the two examples of @xmath21 and @xmath36 , considered in the previous section for @xmath264 , and reconsider them for arbitrary @xmath1 .
we devote the rest of this section to the consideration of @xmath2 , and then turn in the next section to focus exclusively on the case of @xmath3 .
for a polynomial @xmath257 it is trivial that the roots are related to the coefficients by @xmath258 and similar identities for @xmath259 .
these identities can be recast in a way potentially applicable to more general functions , for example entire functions with taylor series @xmath260 , as relations involving the reciprocals of the roots as follows : suppose @xmath261= @xmath262 , then @xmath263 and therefore in general for any @xmath264 and any @xmath7@xmath265 in this formula note that nothing requires @xmath266 .
the multiplicities , @xmath267 , may be negative integers ( @xmath268 a pole ) or indeed arbitrary complex numbers ( @xmath268 a branch point ) .
we shall continue to use the term `` roots '' , or sometimes `` generalised roots '' , to cover all these possibilities .
note also that while ( [ eq : genrootidv1 ] ) gives a family of identities for @xmath269 at any given @xmath7 , this is of course equivalent to knowing the single identity for @xmath17 at general @xmath7 . for
if we have ( [ eq : genrootidv1 ] ) for all @xmath12 at @xmath7 then for @xmath17 at any @xmath270 in ( [ eq : genrootidv1 ] ) we have@xmath271 and vice - versa . unfortunately , of course , while both sides of ( [ eq : genrootidv1 ] ) now make sense for more general functions , the identity is generally untrue for non - polynomials .
for example , for @xmath272 , @xmath33 has no roots so the rhs of ( [ eq : genrootidv1 ] ) is zero while the lhs is identically 1 at @xmath17 for any @xmath7 .
however , suppose @xmath33 fails ( [ eq : genrootidv1 ] ) at @xmath17 with error function@xmath273 and @xmath274 entire .
then if we set @xmath275 where @xmath276 it follows from the fact that the root - sets of @xmath33 and @xmath277 are identical that in ( [ eq : genrootidv1 ] ) for @xmath278 we have@xmath279 thus , even though @xmath33 fails the generalised root identity ( [ eq : genrootidv1 ] ) , there is a unique ( up to an overall scalar ) , nowhere - zero entire function @xmath280 whose product with @xmath261 yields a function with the same root - set satisfying the identity . in terms of equivalence classes , if we set two functions equivalent , @xmath281 , if and only if there exists an entire , nowhere - zero function , @xmath282 , with @xmath283 such that @xmath284 , then within any equivalence class there is a unique representative which satisfies the generalised root identities ( [ eq : genrootidv1 ] ) .
this can be viewed in a different way by thinking of @xmath7 fixed and @xmath269 successively .
taking @xmath150 for example , if the lhs and rhs of ( [ eq : genrootidv1 ] ) differ by @xmath285 for @xmath17 then we can remove this `` obstruction '' by multiplying @xmath33 by @xmath286 since this leaves the root side undisturbed and contributes the required @xmath285 to the derivative side .
similarly , if the obstruction of the @xmath287 identity at @xmath150 is @xmath223 then multiplying @xmath33 by @xmath288 again leaves the root side of ( [ eq : genrootidv1 ] ) unchanged but contributes precisely the required correction of @xmath223 to the @xmath287 identity , while contributing nothing to any of the other identities with @xmath289 and thus leaving these identities undisturbed . in this way
, working successively through @xmath269 we can correct each obstruction and produce a new function @xmath290 which has the same generalised root - set as @xmath33 and does satisfy the root identities ( [ eq : genrootidv1 ] ) for all @xmath12 .
note that if the error function , @xmath291 , obstructing the @xmath17 identity for @xmath33 is not entire , then the situation becomes more complex . in reality the root identities ( [ eq : genrootidv1 ] )
hold without any need for adjustment for many well - known functions .
we next briefly consider three examples .
in the case of @xmath3 , of course , we do not know the location of the non - trivial roots in advance and so our perspective must change . unlike for @xmath2 , where we calculated values for the root - sides of the generalised root identities by cesaro means and used these to infer information about asymptotic behaviour of @xmath2 from the derivative sides ( e.g. stirling s theorem ) , we now aim to calculate the derivative sides of the generalised root identities for @xmath1 and thereby to investigate the location of the roots of @xmath3 by utilising cesaro methods on the corresponding root sides , noting in particular that these cesaro methods naturally depend critically on the geometric location of summands . in this case , for @xmath3 , it turns out that there is no need to do detailed calculations on the derivative side using the fourier definition ; all that is required is the property ( [ eq : muderiva ] ) of @xmath40 .
recall the euler product formula for @xmath3 , namely @xmath525 which is convergent for @xmath152 .
it follows that @xmath526 and so , in light of property ( [ eq : muderiva ] ) , we have that for @xmath527 the derivative side of the root identity for @xmath3 at @xmath0 is given by @xmath528 this expression is clearly convergent for arbitrary @xmath0 for @xmath527 and so gives an expression for the derivative sides of the root identities for @xmath3 for arbitrary @xmath29 and @xmath527 , which can then be extended also to @xmath529 by unique analytic continuation .
since , for arbitrary @xmath527 the sum in ( [ eq : zetaderivsideeuler2 ] ) converges to some finite value as @xmath530 for any @xmath531 , and since @xmath2 has simple poles at all the non - positive integers , so a corollary of the result in ( [ eq : zetaderivsideeuler2 ] ) is the following : [ [ lemma-4 ] ] lemma 4 : + + + + + + + + _ when @xmath1 the derivative sides of the generalised root identities for _ @xmath3 _ are all identically zero as functions of @xmath24 _ ; _ that is _ _ @xmath532 _
since @xmath3 satisfies the generalised root identities , we should therefore have also @xmath533 we now try to calculate the root sides of the generalised root identities for @xmath3 for the cases @xmath4 and @xmath5 in order to verify ( [ eq : zetarootsideidenticallyzeromunegint ] ) in these cases .
first we consider the trivial roots t. for @xmath323 we get the cesaro sum of @xmath534 placed at the points @xmath535@xmath425 . by dilation invariance , this equals the cesaro sum of @xmath534 placed at the points @xmath536@xmath425 and this is familiar from [ 1 ] as giving @xmath196 . for @xmath537
this becomes instead the cesaro sum of @xmath534 placed at the points @xmath538@xmath425 which thus immediately leads to the value @xmath539 , and so in general to arbitrary @xmath24 to derive ( [ eq : zetarootside_mu0_t ] ) here is deliberately heuristic , for reasons of brevity , but the result & similar deductions for the trivial root contributions when @xmath43 and @xmath6 are easy to deduce rigorously by cesaro means if desired ] @xmath540 next , from the simple pole at @xmath250 we get contribution @xmath541 ( since @xmath314 here ) .
now , for the contribution from the non - trivial roots , nt , in the critical strip we recall the explicit form of the riemann - von mangoldt counting function , @xmath44 , which counts roots @xmath542 with @xmath543 . by karatsuba - korolev
[ 5 ] , this has the explicit form @xmath544 where @xmath545 and @xmath49 is the famous argument of the zeta function ( see e.g. [ 5 ] for formal definition ) , and @xmath546 with @xmath547 being the saw - tooth function which rises linearly from 0 to 1 on each integer interval @xmath548 . clearly for the @xmath42 root identity
we are interested in finding the cesaro limit of @xmath44 as @xmath549 , at least after combining with the corresponding calculation for the roots below the real axis ; i.e. , writing @xmath51 for the parameter tracing our count of roots with negative imaginary parts , we have @xmath550 now , by [ 5 ] we know that @xmath551 , and by either [ 5 ] or gordon - sabeh in [ 6 ] we also know that if we define @xmath552 in the usual way , then @xmath553 so that @xmath554(t)=\frac{s_{1}(t)}{t}\rightarrow0 $ ] and @xmath555 . thus ( [ eq : zetarootside_mu0_nta ] ) reduces to @xmath556 as we have seen , however , in considering this cesaro limit , we need to be adding in 1 for each root counted , not at the root @xmath318 itself but at @xmath557 , and this geometry will be critical in our computations , as will the need to combine @xmath48 and @xmath558 before taking cesaro limits ( in particular we will shortly need this to resolve an apparent paradox between results for @xmath42 and @xmath43 ) . to reflect this geometry we thus write @xmath559 so that @xmath560 and therefore @xmath561 and @xmath562 here we are for the moment assuming that , at least for purposes of calculation of cesaro limits , we may treat the parameters @xmath52 and @xmath53 as running purely along the critical line in enumerating the non - trivial zeros .
we shall discuss this assumption ( which would of course be justified if the riemann hypothesis is true , and may be justified in any case by symmetry of roots around the critical line ) further later .
thus we have @xmath563 and therefore @xmath564 we see that taking care to place the nt roots relative to @xmath24 by considering @xmath52 rather than @xmath46 introduces the extra terms of @xmath565 and @xmath566 ( which cancels the @xmath567 to @xmath568 ) in the constant term .
if we add an equivalent constant - term contribution from @xmath558 , arising from the roots in nt below the real axis , we get a constant term of @xmath569 , precisely as required to cancel the contributions from the trivial roots in ( [ eq : zetarootside_mu0_t ] ) and from the simple pole and leave the root side of the @xmath42 root identity with value @xmath45 as hoped - i.e. @xmath570 in agreement with ( [ eq : zetarootsideidenticallyzeromunegint ] ) .
however , there is a subtlety here and we must be careful . if we consider ( [ eq : zetarootside_mu0_ntd ] ) alone then , recalling the dilation invariance of cesaro limits ( so that @xmath571 leaves limits unchanged ) and the fact that @xmath171 and @xmath52 are both pure cesaro eigenfunctions ( with eigenvalue @xmath325 ) and thus have generalised cesaro limit @xmath45 , it follows that @xmath572 but the term @xmath573 has a pure log - divergence which means that , on its own , it has no generalised cesaro limit . here
, however , we see again the importance of being careful in both the geometric placement of roots and in considering the sum of @xmath48 and @xmath558 before taking the cesaro limit , with @xmath51 critically being treated as independent of @xmath46 .
specifically , if we carefully mimic the calculations for ( [ eq : zetarootside_mu0_ntd ] ) for @xmath558 in terms of @xmath53 instead , taking account of the sign changes in ( [ eq : zetarootside_zdefinitionb ] ) and ( [ eq : zetarootside_zdefinitiond ] ) vis - a - vis the corresponding results for @xmath52 , we end up with the corresponding relation @xmath574 in ( [ eq : zetarootside_mu0_ntc ] ) we thus have , on noting ( [ eq : zetarootside_mu0_nte ] ) and its analogue for @xmath575 that @xmath576 as discussed , on combining with the results for t and for the simple pole , we will thus have the desired result that , for @xmath424 @xmath577 if and only if we have that @xmath578 but this is true in spite of the non - cesaro - convergence of either @xmath301 or @xmath579 in the 1-d cesaro setting because , as a 2-d function , @xmath580 can be ascribed a cesaro limit and this limit is @xmath45 . to see this
we first observe that @xmath581 is a direct eigenfunction of the 2-d cesaro operator , @xmath582 , with eigenvalue @xmath131 , since @xmath583=p_{1d}[\ln z]-p_{1d}[\ln\tilde{z}]=(\ln z-1)-(\ln\tilde{z}-1)=\ln\left(\frac{z}{\tilde{z}}\right)\ ] ] as such it can have a 2-d cesaro limit , although the value of this limit is not directly clear .
but then further note that since @xmath580 and @xmath584 should have the same limit , by symmetry , yet @xmath585 , so this limit must in fact be @xmath45 .
it follows at last that we have verified ( [ eq : zetarootsideidenticallyzeromunegint ] ) and the generalised root identity for @xmath3 in the case @xmath42 by direct computation on the root side from the known form of @xmath44 in ( [ eq : zetarootside_zdefinitiona ] ) - ( [ eq : zetarootside_zdefinitiond ] ) .
this confirms that in this case the cesaro count of all roots ( t , pole and nt ) of @xmath3 is identically @xmath45 for all @xmath24 and has been achieved by showing that @xmath586 in a cesaro sense . )
is generically non - integral that there must be infinitely many non - trivial roots of @xmath3 ] in passing we acknowledge that the above argument regarding ( [ eq:2dcesarolimit_ln ] ) is somewhat heuristic .
this is discussed further in section 4.3 and [ 8 ] , but we do note that , if desired , we could change perspective and instead view ( [ eq:2dcesarolimit_ln ] ) as an _ implication _ of @xmath3 satisfying the @xmath42 root identity ( rather than seeking to use ( [ eq:2dcesarolimit_ln ] ) to verify this ) . this would at least then allow us to legitimately use ( [ eq:2dcesarolimit_ln ] ) again in our next calculations for the @xmath43 and @xmath6 root identities , where it reappears .
we thus turn now to the case of @xmath43 .
we will see that a naive calculation on the root side leads to an apparent paradox unless we are careful in again taking account of the geometric locations of @xmath52 and @xmath53 and of their independence .
but doing so we can simultaneously resolve this paradox and verify that , on the root - side of the @xmath43 root identity for @xmath3 we also have @xmath588 as claimed in ( [ eq : zetarootsideidenticallyzeromunegint ] ) , at least modulo an estimate on @xmath49 which certainly holds conditional on rh . in this case , for the trivial roots t , for @xmath323 we get the cesaro sum of @xmath589 placed at the points @xmath535@xmath425 . by dilation invariance
, this equals the cesaro sum of @xmath589 placed at the points @xmath536@xmath425 and this is familiar from [ 1 ] as giving @xmath590 . for @xmath537
this becomes instead the cesaro sum of @xmath591 placed at the points @xmath538@xmath425 which thus immediately leads to the value @xmath592 , and so in general @xmath593 from the simple pole at @xmath250 we get contribution @xmath594 thus , overall , from the combination of the trivial roots and the simple pole we get a total contribution to the root side of the @xmath43 root identity for @xmath3 of @xmath595 turning now to the contribution from the non - trivial roots we have @xmath596 where here we have invoked the symmetry of nt roots w.r.t .
the critical line to take @xmath597 outside the summation as @xmath598 and then utilised result ( [ eq : zetarootside_mu0_nth ] ) from the previous @xmath42 computation . in these terms , the paradox alluded to earlier amounts to claiming naively that , because nt roots also occur in conjugate pairs symmetrically w.r.t .
the real axis , so for real @xmath24 their contributions should cancel the remaining sum on the rhs in ( [ eq : zetarootside_mu-1_nta ] ) to 0 , thus leaving @xmath599 .
this does not cancel the contribution in ( [ eq : zetarootside_mu-1_t_pole ] ) to @xmath600 and would thus leads to a contradiction of ( [ eq : zetarootsideidenticallyzeromunegint ] ) and the generalised root identity for @xmath3 when @xmath43 .
however this paradox is resolved by recalling that @xmath52 and @xmath53 must be treated as independent ; and since the contributions to the sum in ( [ eq : zetarootside_mu-1_nta ] ) from roots above the real axis and from roots below it are both individually classically divergent , we must work very carefully within a cesaro framework to deduce the true cesaro sum of @xmath601 , rather than rely on naive pairwise cancellations .
when we do this we will find that in fact the contribution from @xmath601 is exactly as required to validate that , overall , on the root side of the @xmath43 root identity for @xmath3 we do have @xmath602 and thus ( [ eq : zetarootsideidenticallyzeromunegint ] ) and the generalised root identities remain true for @xmath43 .
to see this , first note that for roots above the real axis we have the partial sum function of @xmath603 given by @xmath604 and similarly for roots below the axis .
thus , after accounting for the difference in sign of @xmath605 in the two cases , we get in ( [ eq : zetarootside_mu-1_nta ] ) that @xmath606 now we shall adopt the notation that @xmath607 and adapt it identically to define @xmath608 @xmath609 @xmath610 for the functions introduced in ( [ eq : zetarootside_zdefinitiona ] ) . in this notation , in ( [ eq : zetarootside_mu-1_ntb ] ) we get ( on noting that all of @xmath611 and @xmath397 have finite limits as @xmath612 , with @xmath613 ) that @xmath614 and similarly for @xmath615 .
now * ( i ) * suppose we have the estimate that @xmath616 then since , in general , @xmath617(t)=\frac{s_{n+1}(t)}{t}\label{eq : p_sn_formula}\ ] ] it would follow immediately that @xmath618(t)\rightarrow0 $ ] as @xmath549 so that @xmath619 also , using integration by parts and noting @xmath620 by definition , we would have @xmath621(t ) & = & \frac{1}{t}\int_{0}^{t}ts_{0}(t)\textrm{d}t = s_{1}(t)-\frac{1}{t}\int_{0}^{t}s_{1}(t)\textrm{d}t\nonumber \\ & = & s_{1}(t)-p[s_{1}](t)\nonumber \\ & \overset{c}{\rightarrow } & 0\label{eq : zetarootside_mu-1_ntf}\end{aligned}\ ] ] thus , combining ( [ eq : zetarootside_mu-1_nte ] ) and ( [ eq : zetarootside_mu-1_ntf ] ) it follows in ( [ eq : zetarootside_mu-1_ntc ] ) that , conditional on the estimate ( [ eq : s2_estimate ] ) , we have no contribution from the @xmath49-related terms , i.e. @xmath622 , and thus @xmath623 and similarly for @xmath615 .
unfortunately , an unconditional estimate of the form ( [ eq : s2_estimate ] ) is not known for @xmath624 .
we do know , however , that a much stronger estimate for all @xmath625 is in fact known , conditional on the riemann hypothesis ( rh ) , namely ( see titchmarsh in [ 7 , pg 354 ] ) @xmath626 as such , we will proceed here on the assumption of ( [ eq : s2_estimate ] ) and ( [ eq : zetarootside_mu-1_ntg ] ) and thus end up obtaining a calculation of @xmath627 conditonal on ( [ eq : s2_estimate ] ) , which is itself implied by rh .
* ( ii ) * from the definition of @xmath628 we clearly have @xmath629 and @xmath630 thus @xmath631 and in light of ( [ eq : zetarootside_zdefinitiona])-([eq : zetarootside_zdefinitiond ] ) we get @xmath632 in similar fashion , we have @xmath633 it follows , on taking cesaro limits after combining @xmath634 and @xmath635 in ( [ eq : zetarootside_mu-1_ntb ] ) ( and recalling @xmath636 ) , that we have @xmath637\\ \\-\left[\tilde{t}\check{n}_{0}(\tilde{t})-\check{n}_{1}(\tilde{t})\right]\end{array}\right\ } & = & \left\ { \begin{array}{cc } 2\pi\left\ { \left(\frac{(s_{0}-\frac{1}{2})^{2}}{(2\pi)^{2}}\right)\cdot(-i\frac{\pi}{4}-i\frac{\pi}{4})\right\ } \\ \\+\underset{z,\tilde{z}\rightarrow\infty}{clim}\left(-\frac{1}{2}\left(\frac{(s_{0}-\frac{1}{2})^{2}}{(2\pi)}\right)\ln\left(\frac{z}{\tilde{z}}\right)\right)\end{array}\right\ } \nonumber \\ \nonumber \\ & = & -\frac{i}{4}(s_{0}-\frac{1}{2})^{2}\label{eq : zetarootside_mu-1_ntm}\end{aligned}\ ] ] on again invoking the 2-d cesaro argument used before to justify taking @xmath638 as per ( [ eq:2dcesarolimit_ln ] ) . *
( iii ) * this leaves just the terms from @xmath639 and @xmath640 in ( [ eq : zetarootside_mu-1_ntb ] ) and ( [ eq : zetarootside_mu-1_ntg ] ) to resolve in order to complete the calculation of the root side of the @xmath43 root identity . using the definition of @xmath50 in ( [ eq : n(t)definition_c ] )
we note from [ 5 ] that @xmath641 and in fact @xmath642 it follows that , in order to understand the cesaro limit of @xmath643 we need to first calculate @xmath285 , which is given by @xmath644 with the initial terms arising from the obvious expansions of @xmath645 and @xmath646 .
to calculate the last term in ( [ eq : zetarootside_mu-1_nto ] ) , although it is a classical limit , we will use a formal cesaro computation and will likewise use such cesaro methods in all calculations related to @xmath397 in the remainder of this paper . since these methods are formal and non - rigorous ( see further discussion in section 4.3 ) we note immediately , however , that all results related to @xmath397 are in fact rigorously verified , either theoretically or numerically , in appendix 5.3 .
we start with the partial integral and apply a formal taylor series expansion of the integrand : @xmath647 on recalling from [ 1 ] that @xmath648 , in ( [ eq : zetarootside_mu-1_nto ] ) we therefore get that we have @xmath649 and thus , in ( [ eq : zetarootside_mu-1_ntn ] ) , we have @xmath650 this suffices to allow computation of the cesaro limit of @xmath651 , and would in fact suffice to allow computation of the contribution from @xmath397-terms in ( [ eq : zetarootside_mu-1_ntg ] ) and ( [ eq : zetarootside_mu-1_ntb ] ) ( since the constant of integration terms from @xmath640 in the @xmath46 and @xmath51 contributions will cancel ) ; but since we will need @xmath652 in detail for the @xmath6 root identity later , we shall calculate @xmath640 more carefully here anyway . from ( [ eq : n(t)definition_c ] ) we have @xmath653 now @xmath654 and @xmath655 in the final term , on reversing the order of integration we get @xmath656_{0}^{t}\textrm{d}u\\ \\ & = & \left\ { \begin{array}{cc } \int_{0}^{\infty}(\frac{1}{2}-\{u\})\ln\left((u+\frac{1}{4})^{2}+\frac{t^{2}}{4}\right)\textrm{d}u\\ \\-2\int_{0}^{\infty}(\frac{1}{2}-\{u\})\ln(u+\frac{1}{4})\:\textrm{d}u\end{array}\right\ } \end{aligned}\ ] ] now the first of these two integrals can be expressed asymptotically in @xmath46 as @xmath657 and a cesaro argument identical to the one just used in deriving ( [ eq : zetarootside_mu-1_ntp ] ) ( as one would expect ) shows this is equal to @xmath658 while the second of the integrals is a constant which we label @xmath659 , i.e. @xmath660 and do not calculate further at this point .
does indeed have a well - defined cesaro value @xmath661 thus , overall , combining terms we get finally in ( [ eq : zetarootside_mu-1_ntr ] ) that @xmath662 where @xmath663 combining ( [ eq : zetarootside_mu-1_ntq ] ) and ( [ eq : zetarootside_mu-1_ntt ] ) in ( [ eq : zetarootside_mu-1_ntg ] ) we then get @xmath664+(\frac{1}{48}-c_{1})+o(\frac{1}{z})\right\ } \label{eq : zetarootside_mu-1_ntv1}\end{aligned}\ ] ] and similarly @xmath665+(\frac{1}{48}-c_{1})+o(\frac{1}{\tilde{z}})\right\ } \label{eq : zetarootside_mu-1_ntv2}\end{aligned}\ ] ] it follows thus in ( [ eq : zetarootside_mu-1_ntb ] ) that @xmath666 on invoking the same 2-d cesaro argument as before . finally , combining ( [ eq : zetarootside_mu-1_ntg ] ) , ( [ eq : zetarootside_mu-1_ntm ] ) and ( [ eq : zetarootside_mu-1_ntw ] ) in ( [ eq : zetarootside_mu-1_ntb ] ) we conclude that the root side of the @xmath43 root identity for @xmath3 has contribution from nt of @xmath667 and on combining with ( [ eq : zetarootside_mu-1_t_pole ] ) this yields that , overall , @xmath668 that is , we have verified as promised that , at least conditional on estimate ( [ eq : s2_estimate ] ) for @xmath624 , which is itself implied by rh , the root side of the @xmath43 root identity for @xmath3 is indeed identically @xmath600 as claimed in ( [ eq : zetarootsideidenticallyzeromunegint ] ) .
hence we now turn to the case of @xmath6 and again try to perform a careful cesaro calculation of the root side of the @xmath6 root identity for @xmath3 to see whether any insight can be gleaned from the claim in ( [ eq : zetarootsideidenticallyzeromunegint ] ) that this , likewise , must be identically equal to @xmath45 . in this case , for the trivial roots t ,
for @xmath323 we get the cesaro sum of @xmath669 placed at the points @xmath535@xmath425 . by dilation invariance , this equals the cesaro sum of @xmath669 placed at the points @xmath536@xmath425 and this is familiar from [ 1 ] as giving @xmath670 . for @xmath537
this becomes instead the cesaro sum of @xmath671 placed at the points @xmath538@xmath425 which thus immediately leads to the value @xmath672 , and so in general @xmath673 from the simple pole at @xmath250 we get contribution @xmath674 thus , overall , from the combination of the trivial roots and the simple pole we get a total contribution to the root side of the @xmath6 root identity for @xmath3 of @xmath675 turning now to the contribution from the non - trivial roots we have @xmath676 writing @xmath677 the first of these sums becomes @xmath678 where @xmath679 on noting that the cross - term in @xmath680 vanishes because the non - trivial roots occur in mirror pairs either side of the critical line ( and this holds identically for any given @xmath46 so there are no cesaro concerns with this cancellation ) .
the sum in the second term becomes , on noting again that in going from @xmath46 to @xmath681 any nt roots off the critical line occur in mirror pairs : @xmath682 on using our results just computed for the non - trivial roots from the case of the @xmath43 root identity ( specifically ( [ eq : zetarootside_mu-1_ntm ] ) and ( [ eq : zetarootside_mu-1_ntw ] ) ) . and
the sum in the third term becomes @xmath683 thus , overall , in ( [ eq : zetarootside_mu-1_nta ] ) we get , after simplification , @xmath684 now we have @xmath685 on noting @xmath686 ; i.e. @xmath687 and similarly for @xmath688 .
recalling the original expression for @xmath47 it follows in ( [ eq : zetarootside_mu-2_ntc ] ) that we have @xmath689 now * ( i ) * as in the case of @xmath43 , suppose here that we have an estimate @xmath690 which , as before , would certainly follow from the much stronger estimates ( [ eq : titchmarshestimates ] ) if rh is true .
then we would have that @xmath691 so that @xmath692(t)\rightarrow0 $ ] and thus @xmath693 ; and , given the capacity to turn cesaro limits of @xmath694 and @xmath695 into cesaro limits of @xmath696 ( in the same fashion as per the argument in ( [ eq : zetarootside_mu-1_ntf ] ) ) , it would follow once again that there was no contribution in ( [ eq : zetarootside_mu-2_nte ] ) from the @xmath697-terms . as in the case of @xmath43 , no such estimate ( [ eq : s3_estimate ] )
is in fact known unconditionally , but we shall proceed here on the assumption of ( [ eq : s3_estimate ] ) , so that @xmath698 and hence continue on towards a calculation of @xmath699 conditional on ( [ eq : s3_estimate ] ) , which is itself implied by rh . *
( ii ) * from our earlier computation of @xmath700 in ( [ eq : zetarootside_mu-1_nti ] ) , on letting @xmath701 , we have @xmath702 and therefore , on combining ( [ eq : zetarootside_zdefinitionb ] ) , ( [ eq : zetarootside_mu-1_nti ] ) and ( [ eq : zetarootside_mu-2_ntf1 ] ) and simplifying , we get that @xmath703 thus @xmath704\\ \\+\frac{i}{9}\left(\frac{(z-(s_{0}-\frac{1}{2}))^{3}}{(2\pi)^{3}}\right)\end{array}\right\ } \nonumber \\ \label{eq : zetarootside_mu-2_ntf3}\end{aligned}\ ] ] and similarly @xmath705\\
\\-\frac{i}{9}\left(\frac{(\tilde{z}-(s_{0}-\frac{1}{2}))^{3}}{(2\pi)^{3}}\right)\end{array}\right\ } \nonumber \\ \label{eq : zetarootside_mu-2_ntf4}\end{aligned}\ ] ] thus , bearing in mind our usual observations that @xmath706 etc , we have , after omission of these terms and simplification , that @xmath707 on invoking the same 2-d cesaro arguments as before . *
( iii ) * from ( [ eq : zetarootside_mu-1_nts ] ) we have @xmath708 where @xmath659
is defined as per ( [ eq : zetarootside_mu-1_adefn ] ) .
now \(a ) @xmath709
( b ) @xmath710 ( c ) @xmath711 and \(d ) reversing the order of integration we get @xmath712 @xmath713 - 2t\\ \\+4(u+\frac{1}{4})\left[\frac{\pi}{2}-\frac{2(u+\frac{1}{4})}{t}+\ldots\right]\end{array}\right\ } \textrm{d}u\nonumber \\ \nonumber \\ & = & \left\ { \begin{array}{cc } 2\left\ { \int_{0}^{\infty}(\frac{1}{2}-\{u\})\textrm{d}u\right\ } t\ln t\\ \\-(2 + 2\ln2)\left\ { \int_{0}^{\infty}(\frac{1}{2}-\{u\})\textrm{d}u\right\ } t\\ \\+2\pi\left\ { \int_{0}^{\infty}(\frac{1}{2}-\{u\})\cdot(u+\frac{1}{4})\textrm{d}u\right\ } + o(\frac{1}{t})\end{array}\right\ } \label{eq : zetarootside_mu-2_nti4}\end{aligned}\ ] ] now ( d)(i ) we saw before that @xmath714 so that , in cesaro terms , @xmath715 and ( d)(ii ) similarly @xmath716 but , extending lemma 12 in [ 1 ] , we have in fact ( see proof in appendix 5.4 ) that [ [ lemma-6 ] ] lemma 6 : + + + + + + + + _ if _ @xmath717 _ then _ @xmath718 applying this in ( [ eq : zetarootside_mu-2_nti6 ] ) we thus get , after simplification , @xmath719 i.e. in cesaro terms @xmath720 then , combining ( [ eq : zetarootside_mu-2_nti5 ] ) and ( [ eq : zetarootside_mu-2_nti7 ] ) in ( [ eq : zetarootside_mu-2_nti4 ] ) it follows that we have , up to @xmath721 , @xmath722 and therefore , finally , combining ( [ eq : zetarootside_mu-2_nti1 ] ) , ( [ eq : zetarootside_mu-2_nti2 ] ) , ( [ eq : zetarootside_mu-2_nti3 ] ) and ( [ eq : zetarootside_mu-2_nti8 ] ) in ( [ eq : zetarootside_mu-2_nth1 ] ) we get , after simplification , @xmath723 where @xmath724 combining ( [ eq : zetarootside_mu-2_nti9 ] ) with our earlier expressions for @xmath639 and @xmath640 in ( [ eq : zetarootside_mu-1_ntq ] ) and ( [ eq : zetarootside_mu-1_ntt ] )
it follows that we have , in the @xmath397-terms in ( [ eq : zetarootside_mu-2_nte ] ) , that @xmath725 where @xmath726 and similarly for @xmath727 . on taking @xmath728 we thus get contribution from these @xmath397-terms in ( [ eq : zetarootside_mu-2_nte ] ) of @xmath729 on noting @xmath730 in the usual way . finally ,
combining our results from ( [ eq : zetarootside_mu-2_ntf ] ) ( conjectural or conditional ) , ( [ eq : zetarootside_mu-2_ntg ] ) and ( [ eq : zetarootside_mu-2_ntk ] ) in ( [ eq : zetarootside_mu-2_nte ] ) , we obtain in ( [ eq : zetarootside_mu-2_ntc ] ) that we have @xmath731 and then , finally , combining ( [ eq : zetarootside_mu-2_ntl ] ) in turn with ( [ eq : zetarootside_mu-2_t_pole ] ) we find that , at least conditional on estimate ( [ eq : s3_estimate ] ) for @xmath732 , which is itself implied by the rh , the root side of the @xmath6 root identity for @xmath3 is given by @xmath733 from this and the fact that , according to ( [ eq : zetarootsideidenticallyzeromunegint ] ) this should equal @xmath45 identically , we can make the following claim ( modulo the rigour of the preceding computations ) : [ [ result-1 ] ] result 1 : + + + + + + + + + the riemann hypothesis ( rh ) is false - that is , there exist non - trivial zeros of @xmath3 off the critical line @xmath734 in the critical strip . [ [ proof-1 ] ] proof : + + + + + + we argue by contradiction . if rh were true , then \(i ) @xmath735 trivially \(ii ) the conditional results regarding cesaro limits of @xmath624 , @xmath736 , and @xmath737 noted in ( [ eq : zetarootside_mu-2_ntf ] ) all actually do hold , by the explicit estimates , conditional on rh , given in [ 7 , pg 354 ] , and \(iii ) all the nt roots @xmath318 are of the form exactly @xmath738 and so the use of @xmath52 and @xmath53 exactly to parametrise the counting functions @xmath44 and @xmath739 in @xmath134 becomes legitimate , since these really do then become step functions on the critical line .
but then , in ( [ eq : zetarootside_mu-2_ntm ] ) we obtain that @xmath740 which contradicts the requirement , from the fact that @xmath3 satisfies the generalised root identities ( and in particular the @xmath6 root identity ) , that @xmath699 should be identically zero for all @xmath24 as per ( [ eq : zetarootsideidenticallyzeromunegint ] ) . the emphasis in this paper has been on the development of certain methods pertaining to remainder cesaro summation / convergence , and on calculation using these methods . consequently , there has been a conscious de - emphasis on formal rigour and detailed proof . in light of this , and especially of the claim in result 1
that such computations imply the riemann hypothesis is false , we now conclude by trying to clarify which claims are in fact rigorously supported and where there are gaps requiring more detailed justification . * ( 1 ) * the first major issue requiring more rigorous justification is that , in the case of @xmath3 , we have simply asserted that because @xmath3 satisfies the generalised root identity for @xmath17 for all @xmath24 ( modulo the @xmath28 obstruction factor ) , so it must ( like @xmath2 ) continue to satisfy these identities for all @xmath24 for @xmath741 . in part
this reflects the belief that , having verified the @xmath17 identity , so that roughly @xmath742 for arbitrary @xmath24 ( the more rigorous statement including the renormalisation correction and @xmath28 term is ( [ eq : hadamardzeta ] ) ) , so further differentiation by @xmath743should imply that the generalised root identities hold for @xmath3 also at arbitrary @xmath0 , and in particular @xmath741 .
but as well as being heuristic this skates over the possibility that there exist functions which act as obstructions for the root identities for @xmath741 in the same way as the functions @xmath450act as obstructions for the root identities for @xmath12 . since @xmath25 does already satisfy the @xmath17 identity for arbitrary @xmath24 ( and hence for @xmath744 ) , so this is the question of whether , for any given @xmath531 , there exists a function , @xmath745 which fails to satisfy the root identities for @xmath287 but satisfies them all at arbitrary @xmath24 for @xmath746 ( since we need to leave these identities undisturbed ) .
we have not addressed this question at all in this paper although , based on considerable experimentation , we believe that the non - existence of such additional obstructions is a reasonable conjecture ( as well as one that seems quite distinct from rh ) .
another way of thinking of this issue is that we know from [ 1 ] ( e.g. the discussion of removable singularities for discrete cesaro schemes discussed in [ 1 ] , section 4.3 ) that the values obtained by cesaro methods at individual parameter values ( e.g. @xmath747 ) are only reliable if they arise from cesaro analysis in an open region around these values which represents a proper analytic continuation ( e.g. in @xmath0 ) to these regions .
since we have only actually verified the root identities @xmath748 for @xmath3 at the isolated values @xmath12 , we have not in fact even established a starting open region in @xmath134 where the root identities are true , let alone systematically extended to regions encompassing the critical points @xmath747 . in a separate paper ( [ 8 ] ) , however , we address this deficiency by showing numerically that it appears that the generalised root identities for @xmath3 are indeed satisfied for all real @xmath749 , where both @xmath750 and @xmath751 are given by classically convergent expressions ; and furthermore that they appear to continue to be satisfied both for values @xmath752 where cesaro divergences need to be removed on the root side and for values @xmath753 where cesaro divergences not only need to be removed but a further cesaro averaging of the resulting residual partial - sum function applied . for @xmath0
further to the left ( i.e. @xmath754 ) the numerical difficulties in verifying the root identities then become too significant without major effort beyond that attempted in [ 8 ] owing to the need to perform additional applications of the cesaro averaging operator @xmath92 , but nonetheless these numerical results in [ 8 ] seem to us to be strongly suggestive that @xmath3 does indeed satisfy the generalised root identities for all @xmath755 ( and hence all @xmath29 by analytic continuation ) as desired .
in addition to this we would also argue that the fact that we have computed the root sides explicitly when @xmath42 and @xmath43 and thereby demonstrated that @xmath3 does satisfy the generalised root identities at these points ( modulo estimate ( [ eq : s2_estimate ] ) in the case of @xmath43 ) is strongly suggestive on its own that @xmath3 should continue to satisfy these identities at @xmath6 , which is the key to result 1 ; and we believe this especially because it seems to us that the nature of the calculation in the @xmath42 and @xmath43 cases is non - obvious and itself suggestive of the aptness of the cesaro approach . *
( 2 ) * a second major issue is that we have not , in fact , rigorously validated the definition of generalised cesaro convergence along a contour given in definition 1 , i.e. we have not rigorously shown that this definition , and in particular the distinction between removing `` eigenfunctions '' in the geometric variable @xmath125 but then just averaging in arc - length @xmath109 , guarantees correct analytic continuation in the situations in which we have applied it ( e.g. in @xmath7 and @xmath0 on the root - side , @xmath351 , of root identities ) .
the numerous examples in the paper where this is confirmed by calculation ( e.g. the definition of @xmath2 , the root identities for @xmath2 at @xmath1 and for @xmath3 at @xmath42 and @xmath43 ) give strong `` experimental '' evidence that this is so , but no rigorous argument has been given .
briefly , such an argument would likely proceed along the following lines , illustrated here in the context of the simpler case of the root side of the root identities for @xmath21 ( but easily adaptable to the case of @xmath3 ) : \(a ) initially restrict to @xmath7 real so that for @xmath756 we are back in the setting of partial sum functions on @xmath8 handled rigorously in [ 1 ] via regular polynomials , @xmath96 , in @xmath92 .
when @xmath0 is real and @xmath749 the series defining @xmath756 is classically convergent , and when @xmath757 the `` removable '' cesaro eigenfunctions are all in fact naturally functions ( e.g. @xmath758 etc ) in the geometric variable @xmath172 \(b ) thus the rigorous methods of [ 1 ] should establish , for all @xmath7 real , the correct analytic continuation of @xmath756 from @xmath749 to all real @xmath0
. analytic extension of @xmath756 to @xmath7 off the real line ( holding @xmath0 real still initially ) is then uniquely determined and seemingly must involve still `` removing '' the same divergent terms in the variable @xmath52 , which is no longer real but has the same geometric meaning , namely @xmath172 as described in definition 1 \(c ) finally , analytic continuation to all @xmath0 off the real line is then also uniquely determined ( although it is no longer necessarily still interpretable in terms of cesaro convergence if @xmath7 is not real ; note that in this paper we have never actually considered non - real @xmath0 ) . with respect to ( b ) , note that the fact that we remove `` eigenfunctions '' such as @xmath758 in the geometric variable @xmath52 ( rather than , say , @xmath759 in @xmath760 ) means that as @xmath0 varies and causes @xmath441 to pass through positive integer values @xmath761 we do not instantaneously pick up discontinuous contributions to the cesaro limits at these @xmath0-values ( as we would if we were removing `` eigenfunctions '' in @xmath109 ) .
this observation bears directly on the remarks ( ii ) and ( iii ) made in commenting upon definition 1 in section 2.1 , and is especially relevant for why we hope that we can still trust the cesaro evaluations in this paper for @xmath2 and @xmath3 at non - positive integer values without need for correction of removable singularities .
note finally , in passing , that the adaptation of the above scheme of argument to the case of @xmath3 would involve negligible change for handling the trivial roots , and would merely require an adaptation of the approach in [ 1 ] from functions on @xmath8 to functions on the positive and negative imaginary axes for handling the @xmath342 and @xmath762 components of the non - trivial roots .
* ( 3 ) * for @xmath3 in the cases of @xmath747 we have relied on a claim that we can ascribe a 2d cesaro limit of @xmath45 to @xmath763 .
in fact , along with a parameter ( say @xmath764 ) for the trivial roots , @xmath46 , the calculations for @xmath751 at @xmath741 are all really 3d cesaro calculations , but we can treat the trivial root calculations independently and it is only in relation to the claim that @xmath638 that we are obliged to consider the two variables @xmath52 and @xmath53 simultaneously . the argument for this claim ( in ( [ eq:2dcesarolimit_ln ] ) ) is , however , clearly non - rigorous ( particularly since we have not formalised the 2d notion , on a pair of contours , being used ) . nonetheless
, we make the following two observations about this claim here : \(a ) first we reiterate that , as noted in section 4.2.1 , for purposes of our claim re rh this 2d cesaro limit could alternatively be taken as an _ implication _ of the @xmath42 root identity and then applied without caveat to the @xmath43 and @xmath6 cases on that basis , and \(b ) secondly , in [ 8 ] , as part of explicitly considering the cesaro treatment of the non - trivial root contributions to @xmath751 in a neighbourhood of @xmath45 , we provide strong numerical evidence to justify taking @xmath765 that @xmath766 as @xmath0 approaches @xmath45 from below .
moreover we give a precise conjecture as to how this @xmath54 divergence may be arising naturally in the limit as @xmath767 ( essentially from the contributions from @xmath342 and @xmath762 each individually diverging as @xmath767 but in opposite directions so that the overall @xmath317-contribution remains finite ) and likewise give strong numerical evidence that this conjecture is true .
this would validate fully that taking @xmath638 at @xmath42 ( and presumably similarly at @xmath43 and @xmath6 ) is the correct thing to do in order to ensure correct analytic continuation . *
lastly , in order to focus on calculation , in numerous areas throughout the paper we have been somewhat loose , for example in : \(a ) adopting a heuristic approach in applying fourier theory to perform calculations on the derivative side in some examples of the generalised root identities , and indeed even in regards to the fourier definition of @xmath32 on the derivative side of these identities in the first place , \(b ) interpreting integrals , @xmath768 or @xmath769 , in a cesaro way as required ( effectively applying a universal cesaro viewpoint throughout the paper ) , \(c ) utilising taylor series expansions without always being careful regarding their domains of convergence , and \(d ) using formal cesaro methods to obtain the required parts of the asymptotic expansions of @xmath770 and @xmath771 in the calculations of the root sides of the root identities for @xmath3 at @xmath747 . for instance , in analysing @xmath770 and @xmath771 as mentioned in ( d ) , despite the fact that these are all classically well - defined functions and thus should be calculable without recourse to cesaro methods , we instead began by applying a taylor series expansion @xmath772 inside an integral , @xmath773 ( namely the integral term in the definition of @xmath397 ) , where for any fixed @xmath46 this expansion is not classically valid for most of the domain of integration @xmath774 .
we nonetheless proceeded by working formally term by term in descending powers of @xmath46 , using cesaro methods ( specifically lemma 6 ) to evaluate the resulting divergent integrals in @xmath775 which then form the coefficients .
and , for @xmath397 , we thereby ended up with result ( [ eq : zetarootside_mu-1_ntq ] ) , which can readily be extended by keeping the next lower order terms in @xmath46 to give a more detailed expansion such as @xmath776 and similar results for @xmath640 and @xmath652 . notwithstanding this `` looseness , '' however ,
we believe that all the calculational claims in this paper are solid . in particular , in relation to the listed areas ( a ) - ( d ) and how they bear on our calculations for @xmath3 , note that : \(i ) on the derivative side of the root identities for @xmath3 at arbitrary @xmath0
we have only ever relied on property ( [ eq : muderiva ] ) of @xmath40 and not needed to explicitly invoke fourier and distributional arguments as @xmath777 and hence also as @xmath778 via the functional equation ; to handle the @xmath2-style asymptotic behaviour that thereby arises as @xmath778 we adapt our arguments regarding stirling s theorem in section 3.4 to this `` one - sided '' setting ( i.e. stirling asymptotics only as @xmath778 not in both directions ) .
details are omitted here however . ] , and \(ii ) as noted in section 4.2.2 , all the claims regarding @xmath770 and @xmath771 used in calculating the root sides of the root identities for @xmath3 at @xmath747 are in fact true and have been verified both numerically and by alternative rigorous derivation ( see appendix 5.3 ) .
overall , despite the issues ( 1)-(4 ) flagged in this section ( some of which are further addressed in [ 8 ] ) , we believe , both for the variety of reasons mentioned in the discussion in this section and on the basis of the results derived throughout the paper in a range of examples , that the methods and results of this paper are correct .
the final figure below depicts relationships among the concepts and results of this paper , and linkages with other well - known results .
it is included in order to provide a survey of the avenues investigated in the paper , and to illustrate why they may provide a new perspective on certain issues , in particular the way they combine in leading to the claim made in result 1 .
[ [ rigorous - verification - of - formal - cesaro - results - for - deltadelta_1-and - delta_2 ] ] rigorous verification of formal cesaro results for @xmath786 and @xmath771 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ in the course of evaluating the root side of the root identities for @xmath3 when @xmath43 and @xmath6 we used the following results regarding @xmath787 and @xmath771 which we derived using formal cesaro means : we now show that these results are all in fact valid despite their heuristic derivation in the body of the paper .
we start with a proof of the asymptotic expansion for @xmath397 in result ( [ eq : appendixdelta1 ] ) .
where@xmath801 and @xmath802is the first `` periodic bernoulli function '' as defined in section 1.1 . using the fact that @xmath803 for @xmath804 ( except at the integer points when @xmath75 which is a set of measure @xmath45 ) , we can write the integral ( [ eq : appendixdelta7 ] ) as @xmath830 and differentiating the right hand side with respect to @xmath46 , we see that , for fixed @xmath831 the right hand side is maximised when @xmath832 .
this maximum @xmath833 is a constant multiple of @xmath834 .
hence @xmath835 and since @xmath836 is at most a constant multiple of @xmath833 , so we can apply the dominated convergence theorem and see that @xmath825 as @xmath549 .
thus in ( [ eq : appendixdelta8 ] ) we have that the integral ( [ eq : appendixdelta7 ] ) is given by where @xmath842 is euler s constant .
but since @xmath843 it follows easily that @xmath844(k+\alpha)\rightarrow0 $ ] as @xmath487 .
hence we have that the partial integral @xmath845 does have a generalised cesaro limit as @xmath487 and in fact we have this leaves just the verification of the asymptotic expressions ( [ eq : appendixdelta2 ] ) and ( [ eq : appendixdelta3 ] ) for @xmath847 and @xmath771 .
fully rigorous derivations of these have also been obtained along the lines of the derivation above for @xmath397 , after obtaining recurrence relations and certain preliminary estimates for the integrals @xmath848 for the sake of brevity , however , we omit details here . instead , as an alternative we note that the expressions in ( [ eq : appendixdelta1])-([eq : appendixdelta3 ] ) for @xmath849 and @xmath771 can also all be readily verified numerically by writing the integral in the definition of @xmath397 as a convergent sum over @xmath180 of integrals from @xmath180 to @xmath850 ; since on each such unit interval the integral is easy to perform exactly ( and likewise for the furrther integrations for @xmath847 and @xmath771 ) , so the resulting expressions for @xmath849 and @xmath771 are comprised entirely of elementary functions and convergent sums of the same , allowing thorough numerical verification of the claimed asymptotic expansions .
such sample numerical confirmations , implemented via vba code , can be found in the xl2007 spreadsheet `` delta_0_1_2_asymptoticsfinal2.xlsm '' made available with this paper . in these cases the specific claims required ( that @xmath852 , @xmath853 , @xmath854 , and @xmath855 ) all involve only @xmath856 or @xmath857 and are readily checked independently by direct calculation . however , since we regard lemma 6 as an independently interesting result in the cesaro framework , we include its general proof here now . on the one hand , if @xmath863 for all @xmath118 then @xmath864 on recognising the binomial expansion of @xmath865 . on the other , if @xmath866 for all @xmath118 then @xmath867 on using the fact that only the @xmath868 term is non - zero and invoking ( [ eq : fg_n0_r ] ) .
thus the result of lemma 7 holds in both directions .
@xmath871(x ) & = & \frac{1}{x}\int_{0}^{k+\alpha}f_{1,r}(t)\,\textrm{d}t\\ \\ & = & \frac{1}{x}\left\ { \sum_{j=0}^{k-1}\frac{j}{r+1}+k\cdot\frac{\alpha^{r+1}}{r+1}\right\ } \\ \\ & = & \frac{1}{x(r+1)}\left\ { \frac{1}{2}(x-\alpha)(x-\alpha-1)+(x-\alpha)\alpha^{r+1}\right\ } \\ \\ & \overset{c}{\sim } & \frac{1}{2(r+1)}\left\ { x-(2\alpha+1)+2\alpha^{r+1}\right\ } + o(1)\\ \\ & \overset{c}{\rightarrow } & \frac{1}{2(r+1)}\left\ { -2+\frac{2}{r+2}\right\ } = -\frac{1}{r+2}\end{aligned}\ ] ] on invoking ( [ eq : fg_n0_r ] ) and recalling that @xmath80 is an eigenfunction of @xmath92 with eigenvalue @xmath325 so that @xmath872 .
this verifies lemma 6 and the formulae in ( [ eq : fg_lemma7 ] ) when @xmath857 as our base case .
@xmath875(x ) & = & \frac{1}{x}\left\ { \sum_{j=0}^{k-1}\frac{j^{n}}{r+1}+k^{n}\cdot\frac{\alpha^{r+1}}{r+1}\right\ } \nonumber \\ \nonumber \\ & = & \frac{1}{x(r+1)}\left\ { b_{n+1}(k-1)+k^{n}\cdot\alpha^{r+1}\right\ } \label{eq : fg_gen_n1}\end{aligned}\ ] ] if @xmath881 the inductive hypothesis applies to all except the @xmath882 term , and when @xmath883 it applies to all except the @xmath882 and @xmath884 terms . in the last case of @xmath885
, however , the resulting function is a pure power of @xmath80 with generalised cesaro limit
@xmath45 in any case , and so the only contributions across the board come from the @xmath882 terms , namely @xmath887(x)\overset{c}{\rightarrow}\frac{(-1)^{n}}{(r+1)}+\frac{(-1)^{n-1}n}{(r+1)(n+r+1)}=\frac{(-1)^{n}}{(n+r+1)}\ ] ] and this completes the inductive step , thus completing the proof of lemma 6 and the equivalent formulae in lemma 7 . | we extend the notion of generalised cesaro summation / convergence developed in [ 1 ] to the more natural setting of what we call `` remainder '' cesaro summation / convergence and , after illustrating the utility of this approach in deriving certain classical results , use it to develop a notion of generalised root identities .
these extend elementary root identities for polynomials both to more general functions and to a family of identities parametrised by a complex parameter @xmath0 . in
so doing they equate one expression ( the derivative side ) which is defined via fourier theory , with another ( the root side ) which is defined via remainder cesaro summation . for @xmath1 these identities
are naturally adapted to investigating the asymptotic behaviour of the given function and the geometric distribution of its roots . for the gamma function
we show that @xmath2 satisfies the generalised root identities and use them to constructively deduce stirling s theorem . for the riemann zeta function , @xmath3 ,
the implications of the generalised root identities for @xmath4 and @xmath5 are explored in detail ; in the case of @xmath6 a symmetry of the non - trivial roots is broken and allows us to conclude , after detailed computation , that the riemann hypothesis must be false . in light of this
, some final direct discussion is given of areas where the arguments used throughout the paper are deficient in rigour and require more detailed justification .
the conclusion of section 1 gives guidance on the most direct route through the paper to the claim regarding the riemann hypothesis . |
the photoionization process is one of the important radiative feedback processes in astrophysics @xcite .
the increase in pressure caused by photoionization can trigger strong dynamic effects , such as photoionization hydrodynamics .
the challenge in combining hydrodynamics with photoionization lies in the difference in time scales between the two process .
photoionization ( pi ) and photoabsorption ( pa ) processes play important roles in many physical systems , including a broad range of astrophysical objects as diverse as quasi - stellar objects ( qsos ) , the atmosphere of hot stars , proto - planetary nebula , hii regions , novae and supernovae .
chandra _ and _ xmm - newton _ satellites currently provides an abundance of x - ray spectra from astronomical objects ; high - quality atomic data is needed to interpret such spectra @xcite .
theoretical studies , recently made on atomic carbon and it ions indicated high quality atomic data are necessary to accurately model _ chandra _ observations in the x - ray spectrum of the blazar mkn 421@xcite . in the soft x - ray region
( 5 - 45 ) , spectroscopy , including _
k_-shell transitions for atomic elements such as , c , n , o , ne , s and si , in neutral , or low stages of ionization and _ l_-shell transitions of fe and ni , are a valuable tool for probing the extreme environments in active galactic nuclei ( agn s ) , x - ray binary systems , cataclysmic variable stars ( cv s ) and wolf - rayet stars @xcite , and the ism @xcite . interstellar oxygen is found in both gas and dust phases , although the exact molecular form of the dust remains unknown .
therefore an accurate understanding of the gas phase constituents is necessary in order to measure the residual molecular and solid - phase components @xcite . in the x - ray community , electron - beam - ion - trap ( ebit ) measurements ( used for calibrating resonance energies ) ,
have been carried out for the inner - shell 1s @xmath7 2@xmath8 transitions in he - like and li - like nitrogen ions @xcite , li - like , be - like , b - like and c - like oxygen ions @xcite . in ebit experiments ,
the spectrum is contaminated and blended with ions in multiple stages of ionization , making spectral interpretation fraught with difficulties .
cleaner , higher - resolution spectra are obtained at synchrotron radiation facilities ; als , bessy ii , soleil , astrid ii and petra iii .
ebit experiments have the advantage of the production of pure ground state populations of atoms or ions , extremely difficult to make with merged beams methods , routinely used at synchrotron radiation facilities .
photoionization and photoabsorption cross sections used for the modelling of astrophysical phenomena has traditionally been provided by theory , as limited experimental data is available across a wide range of wavelengths . until recently ,
the bulk of theoretical work has not been tested thoroughly by experiment @xcite . for atomic oxygen
the availability of x - ray data on this system provided the motivation to perform theoretical _
k_-shell photoionization investigations .
inner - shell excitation processes occurring with the interaction of a photon on the @xmath9 ground - state of atomic oxygen produces strong resonances observed in the corresponding cross section ( c.f .
, figure 2 and table 1 ) , through promotion of the 1s @xmath7 @xmath10 electron via the processes ; @xmath11 } np ~{\rm ^3p}^{\circ})\ ] ] giving competing decay routes namely , @xmath12 ) + e^- ( { \it k^2_{\ell}}),}\ ] ] and @xmath13 ) } n{^{\prime}}{\ell^{\prime } } + e^- ( { \it k^2_{\ell}),}\ ] ] which the present theoretical approach attempts to simulate , where , @xmath14 = 2 6 ( observed in the experiment ) , and @xmath15 is the outgoing energy of the continuum electron with angular momentum @xmath8 . _ k_-shell photoionization contributes to the ionization balance in a more complicated way than outer shell photoionization . _ k_-shell photoionization when followed by auger decay couples three or more ionization stages instead of two in the usual equations of ionization equilibrium @xcite . early theoretical photoionization ( pi ) cross section calculations for _ k_-shell processes on this complex performed by reilman and manson @xcite used the hartree - slater wavefunctions of herman and skillman @xcite and the dirac - slater wavefunctions @xcite .
photoionization cross sections determined from central field approximations although excellent for high photon energies yield unreliable results near thresholds ( see figure 3 ( a ) ) , where resonance features dominate the cross sections @xcite .
state - of - the - art _ ab initio _
calculations for photoabsorption cross - sections and auger inner - shell processes were first investigated on this system , using the standard r - matrix approach @xcite , for modelling the resolved interstellar o _ k _ , ne _ k _ , and fe _ l_-edge absorption spectra in the _ chandra _ x - ray observatory low - energy transmission grating spectrometer ( letgs ) spectrum of the low - mass x - ray binary x0614 + 091 @xcite .
this work was extended using the optical potential technique @xcite to account for auger broadening of resonances below the @xmath0 threshold and to analyze the high - resolution spectroscopy of the oxygen _ k_-shell interstellar absorption edge in seven x - ray binaries @xcite .
garcia and co - workers @xcite , using the optical potential method within the breit - pauli r - matrix formalism @xcite extended this work to the oxygen iso - nuclear sequence and to investigate the x - ray absorption structure of atomic oxygen in the interstellar medium by analyzing _
xmm - newton _ observations of the low - mass x - ray binary sco x-1 @xcite .
cross sections for atomic oxygen _
k_-shell photoionization were measured over the photon range 520 ev to 555 ev .
our new results were obtained on undulator beamline 11.0.2 at the als , previous measurements were performed on bend - magnet beamline 6.3.2 @xcite .
the present experimental measurements have covered the complete k - shell region , in a single scan , rather than constructing the spectra piece - meal like as was done previously @xcite . a resolution of 4250 @xmath3 400 ( @xmath2 124 @xmath3 12 mev ) at 526.8 ev , and a photon flux of over @xmath16 photons / s
was provided by bl 11.0.2 when using slit widths of 20 @xmath17 m .
the experimental apparatus has been discussed previously in detail @xcite .
( colour online ) photoionization cross sections of o@xmath18 , o@xmath19 and o@xmath20 produced by the decay of a @xmath21 hole in atomic oxygen .
the resonance lines represent the transitions @xmath22 , @xmath23 and @xmath24 with @xmath14 = 3 - 6 .
the current resolving power of the monochromator was 4250 @xmath3 400 ( @xmath2 124 @xmath3 12 mev ) at a photon energy of 526 ev.,width=302 ] similar to our earlier results @xcite , we used the rydberg resonance features found in molecular oxygen near 541 ev to calibrate the photon energy scale , resulting in a maximum uncertainly of 40 mev . during double - bunch operations of the als , a wiley - mclaren style time - of - flight mass spectrometer @xcite , oriented with its axis parallel to the polarization vector of the incident synchrotron radiation ,
was used in conjunction with a microwave discharge system to determine the branching ratios for atomic oxygen at the @xmath25 resonance .
both the microwave _ on _ ( o and o@xmath26 mixture ) and _ off _ ( o@xmath26 ) independent spectral scans were divided by the incident photon flux .
the molecular contribution to the microwave _ on _ spectrum was removed by a scaled subtraction of a microwave _ off _ spectrum at the o@xmath26 : 1@xmath27 resonance ( 530.5 ev ) .
several residual peaks are created due to the molecular resonances being a few mev wider for the microwave _ on _ spectra .
we surmise this difference is probably due to the larger thermal motion of the gas molecules with the microwave discharge on .
the scans were finally corrected by removal of the background produced by direct photoionization of the atomic oxygen valence shell .
this was only necessary for o@xmath18 or o@xmath19 , being that o@xmath20 sat on a zero background .
we performed two coarse photon energy scans , one for each ion , covering the range between 500 600 ev .
the o@xmath18 or o@xmath19 signal produced by _
k_-shell photoionization is superimposed on a nearly flat background caused by direct photoionization of the valence shell .
this background contributed 35% and 8% to the total o@xmath18 and o@xmath19 signals , respectively , just above the 1s@xmath28 @xmath29p series limit .
the independent ion specific photon energy scans could then be placed onto a relative scale by using the branching ratios ( o@xmath18/tiy = 80.07% , o@xmath19/tiy = 17.32% , and o@xmath20/tiy = 0.026% , with tiy = total ion yield ) measured with the time - of - flight mass analyzer on top of the @xmath30 resonance .
finally , the scans were placed onto an absolute scale ( see figure 1 ) by summing their values above the @xmath29p ionization limit and normalizing this sum to the difference of the cross sections above and below the oxygen _ k_-edge @xcite .
the @xmath31-matrix with pseudo - states method @xcite was used to determine our theoretical results .
cross section calculations were performed in @xmath32-coupling retaining 910-levels ( valence and hole - states ) of the residual o@xmath18 ion in the close - coupling expansion .
hartree - fock @xmath33 , @xmath34 and @xmath35 @xcite and n=3 pseudo - orbitals of the o@xmath18 residual ion ( included for core relaxation and correlations effects ) were used , obtained by optimizing on the energy of the hole state ; @xmath36 , @xmath37 and @xmath38 on @xmath39 including the important configuration @xmath40 with the multi - configuration - hartree - fock ( mchf ) atomic structure code @xcite . for the o(@xmath41 @xmath42p ) bound state we obtained 14.0344 ev for the ionization potential using a triple electron promotion model , the nist experimental value is 13.61806 ev , a discrepancy of @xmath43 3 % , or 416 mev . a double electron promotion model gave 13.82184 ev , a discrepancy of @xmath43 1.5 % , or 204 mev , yielding closer agreement with the nist tabulated value as the rmps approach provides more highly correlated wave functions . in previous work @xcite obtained an underestimate of the ionization potential @xmath43 2.5 % , or 338 mev compared to experiment , due to limited correlation included . in the collision calculations for atomic oxygen , twenty continuum functions and a boundary radius of 7.27 bohr radii
two and three - electron promotion scattering models were investigated giving similar results . the collision problem was solved using an energy grid of 2@xmath4410@xmath45 rydbergs ( @xmath2 2.72 @xmath17ev ) allowing detailed resolution of resonance features in the cross sections .
the peaks found in the photoabsorption cross section spectrum were fitted to fano profiles @xcite instead of the energy derivative of the eigenphase sum technique @xcite .
theoretical values for the natural linewidths @xmath4 ( mev ) are presented in table 1 and compared with current and previous als measurements @xcite and with prior investigations .
( colour online ) atomic oxygen photoabsorption cross sections taken at 124 mev fwhm compared with theoretical estimates .
the r - matrix calculations shown are from the rmps method ( solid black line , present results ) convoluted with a gaussian profile of 124 mev fwmh .
table 1 designates the resonances and their properties.,width=321 ]
in the photon energy range ( 520 ev 555 ev ) explored , intense structure is observed in the cross section between 520 ev and 530 ev , from the strong @xmath46 transition in the atomic oxygen spectrum .
figure 2 shows our present experimental and theoretical results for the photon energy range of 520 ev 555 ev illustrating all the additional @xmath47 transitions ( n @xmath48 3 ) in the spectrum .
previous experimental measurements @xcite were actually measured at a photon energy resolution of 135 mev and not 182 mev , current als measurements are at 124 mev fwhm .
convolution of the rmps theoretical results with a gaussian function of 124 mev fwhm was used to compare directly with the als measurements .
resonances observed in the experimental measurements were fitted with voigt profiles to determine the natural linewidths using a gaussian function of 124 mev fwhm for each peak .
the photon energy was calibrated to an energy uncertainty of approximately @xmath3 40 mev .
previous measurements of stolte and co - workers @xcite were re - analyzed with the present results .
the measured spectra for o@xmath49 , o@xmath19 and o@xmath20 production were fitted with the program winxas@xmath50 of thorsten ressler , hamburg , germany @xcite and its near edge x - ray absorption fitting routines . due to an incomplete data set , previous measurements for o@xmath20
@xcite were not fitted . in previous measurements
@xcite , a width of 231 mev was cited with a resolution of 182 mev .
on refitting the previous results @xcite it was discovered that it was not possible to arrive at a proper fit with a resolution of 3000 .
current multi - function fits , using voigt and arctan functions , determined the resolution to be 3800 @xmath3 150 ( @xmath2 135 @xmath3 5 mev ) .
the 1s2s@xmath292p@xmath51(@xmath51p@xmath52)6p and 1s2s@xmath292p@xmath51(@xmath29p@xmath52)6p states can not be properly fitted , since they are completely hidden .
rydberg s formula was used to determine the resonance energies , given by , @xmath53 where , @xmath54 is the resonance transition energy , in rydbergs , @xmath55 the ionization potential and the resonance series limit . the principal quantum number @xmath14 , the effective quantum number @xmath56 and the quantum defect @xmath17 are related by @xmath56 = @xmath57 @xcite . converting all quantities to ev ,
members of the rydberg series are represented by , @xmath58 @xmath59 is the resonance energy position , @xmath60 the ionisation limit , @xmath61 is the charge of the core ( in this case , @xmath62 = 1 ) and @xmath63 is 13.6057 ev @xcite . in figure 2
we present the experimental cross section measurements from the als taken at 4250 @xmath3 400 ( @xmath2 124 @xmath3 12 mev ) resolution compared to our theoretical work . in the non - resonant region , above the k edge , at 550 ev , theory gives a value of 0.560 mb , the als experimental measured value is 0.559 mb , a discrepancy of 0.03% .
figure 2 shows the excellent agreement between theory and experiment over the entire energy region and figure 3 ( a ) illustrates the rmps results with the central field approximation @xcite results .
strong resonance features near the k - edge present in the state - of - the - art rmps cross sections are absent from the central field calculations . in figure 3(b ) un - convoluted , rmps cross sections with the optical potential r - matrix results @xcite are presented .
note , the discontinuity ( @xmath43 538 ev ) in the optical potential r - matrix cross section results absent from the rmps calculations . finally , figure 4 compares quantum defects @xmath17 obtained from the rmps and the optical potential methods @xcite with experiment .
table 1 presents our experimental and theoretical results , for resonance energies , resonance strengths @xmath64 , quantum defects @xmath17 and natural linewidths @xmath65 with previous theoretical and experimental work @xcite .
table 1 includes the experimental values for the lifetime @xmath66 expressed in femto - seconds ( fs ) , determined via the uncertainty principle ( @xmath67 = @xmath68/2 ) .
_ k_-shell photoabsorption of atomic oxygen was investigated using the r - matrix with pseudo - states ( rmps ) method along with current and previous high resolution experimental measurements made at the als .
resonance features observed in the cross sections for photon energies in the range 520 ev
555 ev , are identified as 1s @xmath7 np transitions that are analyzed , natural linewidths @xmath4 ( mev ) extracted and lifetimes determined via the uncertainty principle .
excellent agreement ( see table 1 ) of theoretical estimates , for the resonance parameters ( resonance energies , natural line widths and quantum defects ) compared to experimental measurements is obtained with calculations for the ionization potential accurate to within 1.5 % of experiment .
we have delineated _ all _ of the resonances properties and made a detailed comparison of experiment with current and previous theoretical investigations .
earlier theoretical work @xcite made a limited comparison with experiment , for the quantum defects of the two resonance series with only the auger width of @xmath69 state determined .
the present als high resolution ( 124 mev fwhm ) of all the resonance peaks observed in the cross sections over the _ entire _ energy range investigated allowed a direct comparison with state - of - the - art r - matrix calculations to be made .
the only limitation of the present theoretical model is apparent from table 1 concerning the resonance strengths which might be addressed in the future with an extended pseudo - states basis within a breit - pauli approximation . finally , the data is suitable to be incorporated into the astrophysical modelling codes cloudy @xcite , xstar @xcite and atomdb @xcite .
( colour online ) ( a ) r - matrix ( rmps ) cross sections ( convoluted at 124 mev ) compared with central field approximation results @xcite , ( b ) cross sections ( unconvoluted ) from the rmps ( solid black line ) and optical potential r - matrix calculations ( solid red line).,width=321 ] the energy resolution on present day satellites , _
chandra _ and _ xmm - newton _ , is @xmath43 0.6 ev , a factor of 10 lower than available at current ground based synchrotron radiation facilities , like the als , soleil , astrid ii , bessy ii or petra iii , providing higher resolution and precision than obtained via satellites .
there are also issues concerning the calibration of spectra obtained from satellites as previously highlighted .
the als measurements have been calibrated to the resonance transition in molecular oxygen , which is known very accurately .
cccccccc @xmath70 & 526.83 & 526.79 @xmath3 0.04 & 1.110 & 1.110 @xmath3 0.003 & 150@xmath71 & 148 @xmath3 11@xmath72 & 2.22
@xmath3 0.17 + & & & & & 169@xmath52 & 153 @xmath3 12@xmath73 & - + & & & & & 139@xmath74 & 160 @xmath3 9 @xmath75 & - + @xmath76 & 541.23 & 541.19 @xmath3 0.04 & 0.796 & 0.811 @xmath3 0.023 & 143 & 167 @xmath3 11 & 1.97 @xmath3 0.13 + @xmath77 & 542.70 & 542.68 @xmath3 0.04 & 0.801 & 0.825 @xmath3 0.063 & 97 & 125 @xmath3 11 & 2.63 @xmath3 0.23 + @xmath78 & 543.25 & 543.23 @xmath3 0.04 & 0.824 & ( 0.800 @xmath3 0.29 ) & 70 & 119 @xmath3 10 & 2.77 @xmath3 0.23 + @xmath79 & 543.52 & 543.51 @xmath3 0.04 & 0.833 & ( 0.800 @xmath3 0.29 ) & 80 & 196 @xmath3 12 & 1.68 @xmath3 0.10 + - & - & - & - & - & - & - & - + - & - & - & - & - & - & - & - + @xmath80 & 544.03 & 544.03 @xmath3 0.04 & & & & & + + @xmath76 & 545.97 & 545.83 @xmath3 0.04 & 0.826 & 0.870 @xmath3 0.015 & 187 & 196 @xmath3 10 & 1.68 @xmath3 0.10 + @xmath77 & 547.48 & 547.45 @xmath3 0.04 & 0.848 & 0.871 @xmath3 0.045 & 127 & 128 @xmath3 10 & 2.57 @xmath3 0.20 + @xmath78 & 548.04 & 548.04 @xmath3 0.04 & 0.902 & ( 0.870 @xmath3 0.105 ) & 102 & 156 @xmath3 10 & 2.11 @xmath3 0.14 + @xmath79 & 548.35 & 548.33 @xmath3 0.04 & 0.784 &
( 0.870 @xmath3 0.201 ) & 110 & 157 @xmath3 13 & 2.10 @xmath3 0.17 + - & - & - & - & - & - & - & - + - & - & - & - & - & - & - & - + @xmath81 & 548.85 & 548.85 @xmath3 0.04 & & & & & + & & & & & & & + & & & strengths & strengths & & & + & & & theory & experiment & & & + & resonance & & ( r - matrix@xmath71 ) & ( als@xmath72 ) & & & + & & & & & & & + & series & & @xmath82&@xmath82 & & & + & @xmath83 & & ( mb ev ) & ( mb ev ) & & & + & @xmath84 & & 10.22 & 5.76 @xmath3 1.45 & & & + & @xmath85 & & 0.88 & 0.68 @xmath3 0.17 & & & + & @xmath86 & & 0.30 & 0.25 @xmath3 0.06 & & & + & @xmath87 & & 0.13 & 0.08 @xmath3 0.02 & & & + & @xmath88 & & 0.09 & - & & & + + & @xmath89 & & & & & & + & @xmath90 & & 1.12 & 0.58 @xmath3 0.15 & & & + & @xmath86 & & 0.44 & 0.17 @xmath3 0.04 & & & + & @xmath91 & & 0.20 & 0.07 @xmath3 0.02 & & & + & @xmath88 & & 0.16 & - & & & + + in contrast to this , observed spectra using the letg on _ chandra _ , are calibrated to either theoretical calculations or ebit measurements @xcite for the @xmath92 resonance line in h - like , he - like oxygen , li - like or b - like oxygens ions ( with all the ensuing complications and difficulties of ion identification ) .
given the high precision of our experimental and theoretical data we recommend that observational data concerning _ k_-shell photoabsorption of atomic oxygen be calibrated to the present work .
bmmcl , and cpb thank the institute for theoretical atomic and molecular physics ( itamp ) , at the harvard - smithsonian center for astrophysics for their hospitality and support ( bmmcl ) under the visitor s program .
itamp is supported by a grant from the national science foundation .
cpb acknowledges support by us department of energy ( doe ) grants through auburn university .
wcs acknowledges support from the national science foundation under nsf grant no .
phy-01 - 40375 .
we thank professor alex dalgarno frs , dr john c raymond , dr randall k smith , dr jeremy j drake and dr brad wargelin for discussions on the astrophysical applications and the letg _ chandra _ calibration .
grants of computational time at the national energy research scientific computing center in oakland , ca , usa , the kraken xt5 facility at the national institute for computational science ( nics ) in knoxville , tn , usa and at the high performance computing center stuttgart ( hlrs ) of the university of stuttgart are gratefully acknowledged .
the kraken xt5 facility is a resource of the extreme science and engineering discovery environment ( xsede ) , which is supported by national science foundation grant no .
the help of hendrik blum and tolek tyliszczak in setting up the experiment on beamline 11 at the als is gratefully acknowledged . the advanced light source in berkeley , ca
, usa , is supported by the director , office of science , office of basic energy sciences , of the us department of energy under contract no .
de - ac02 - 05ch11231 . | photoabsorption of atomic oxygen in the energy region below the @xmath0 threshold in x - ray spectroscopy from _ chandra _ and _ xmm - newton _ is observed in a variety of x - ray binary spectra .
photoabsorption cross sections determined from an r - matrix method with pseudo - states ( rmps ) and new , high precision measurements from the advanced light source ( als ) are presented .
high - resolution spectroscopy with e/@xmath1e @xmath2 4,250 @xmath3 400 was obtained for photon energies from 520 ev to 555 ev at an energy resolution of 124 @xmath3 12 mev fwhm .
_ k_-shell photoabsorption cross - section measurements were made with a re - analysis of previous experimental data on atomic oxygen at the als .
natural linewidths @xmath4 are extracted for the @xmath5 and @xmath6 rydberg resonances series and compared with theoretical predictions .
accurate cross sections and linewidths are obtained for applications in x - ray astronomy .
excellent agreement between theory and the als measurements is shown which will have profound implications for the modelling of x - ray spectra and spectral diagnostics . |
x - rays serve as one of our primary probes of magnetic activity in solar - type stars . thus
, much of our understanding of key physical processes thought to be connected to magnetic fields
stellar winds , for example derives from the study of stellar x - rays .
one of the most compelling stories in stellar astrophysics told through x - rays is that of the intimate relationship between stellar rotation and magnetic field generation . indeed , among late - type main - sequence stars , rotation is the strongest correlate of x - ray luminosity , and the observed rotation / x - ray relationship @xcite has become central to the current paradigm of dynamo - generated magnetic fields , of magnetically driven stellar winds , and ultimately of the evolution of stellar angular momentum .
the observed relationship between rotation and x - ray emission on the main sequence is remarkably clean , and clearly separates stars into three regimes ( cf .
@xcite ) typically described phenomenologically as the linear , saturated , and super - saturated regimes , in order of increasing stellar rotation @xcite . for slowly rotating stars
the x - ray luminosity , @xmath3 , scales linearly with the stellar angular velocity , @xmath4 , consistent with the theoretical idea that more rapid stellar rotation produces a stronger magnetic field through an @xmath5-type dynamo ( for stars with radiative cores and convective envelopes ) or through a distributed turbulent dynamo ( for fully convective stars ) . for stars rotating more rapidly than a certain threshold ,
the x - ray luminosity is observed to saturate " at a fixed value relative to the stellar bolometric luminosity : @xmath6 . while the reasons for saturation have not been well understood ,
this observation has had important ramifications for efforts to model the angular momentum evolution of young solar - type stars @xcite .
the models now routinely include saturation as a key ingredient in their parametrizations of angular momentum evolution through winds . finally , the most rapidly rotating stars exhibit x - ray emission at levels roughly a factor of 2 below the saturation value .
this super - saturation " effect @xcite has been poorly understood .
@xcite has recently re - interpreted super - saturation in terms of a new paradigm for the angular momentum evolution of solar - type stars ; whether this new paradigm will survive detailed scrutiny is not yet clear .
what is clear is that the rotation / x - ray relationship serves as a key observational touchstone for developing and evaluating our theoretical understanding of the generation and evolution of stellar magnetic fields , the generation and evolution of stellar winds , and the evolution of stellar angular momentum .
considerable observational effort has been invested , therefore , in trying to establish the presence of a rotation / x - ray relationship among pre
main - sequence ( pms ) stars , where the questions of magnetic field generation and evolution , winds , and angular momentum evolution remain largely unanswered . unfortunately , to date these efforts have not borne much fruit .
in one recent study of t tauri stars ( tts ) in taurus - auriga @xcite , a rotation / x - ray correlation has been reported , but the sample size is small ( @xmath7 ) and there are lingering concerns with respect to completeness / reliability of the rotation periods and biases in the sample both astrophysical and observational . in particular , as discussed by @xcite , the taurus - auriga pms population appears to be deficient in both high - mass stars and faint , low - mass weak - lined tts .
in addition , rotation periods in this region were determined from photometric monitoring campaigns some 10 years ago that were relatively sparsely sampled and of relatively short duration , and therefore potentially biased against both very fast and slow rotators .
two new studies based on deep _ chandra _ observations of the orion nebula cluster ( onc ) provide the most comprehensive analyses yet of x - rays and rotation in a large , coeval ( @xmath8 myr ) sample of pms stars .
neither the study of flaccomio and collaborators @xcite with the high resolution camera ( hrc ) , nor that of feigelson and collaborators @xcite with the advanced ccd imaging spectrometer ( acis ) , found evidence for a rotation / x - ray relationship such as that observed on the main sequence or that reported for tau - aur tts by @xcite . indeed
, these studies find that it is stellar mass that is by far the dominant correlate of pms x - ray luminosity , with @xmath9 correlating with rotation either not at all , or perhaps slightly in the opposite sense from the main - sequence rotation / x - ray relationship .
both studies found that stars with known rotation periods have x - ray luminosities near the main - sequence saturation value of @xmath6 .
moreover , both studies suggest that pms stars might in fact be _ expected _ to reside in the super - saturated regime , considering that typical pms rossby numbers @xcite are small due to the long convective turnover times ( @xmath10 days ) of these very young and fully convective stars . thus , while direct observation of pms stars in the linear regime of the rotation / x - ray relationship remains elusive , these studies seem to confirm , if indirectly , the basic picture of the rotation / x - ray relationship by suggesting that all onc stars are in the super - saturated regime , and that this is where they ought to be . but not all onc stars detected by _
chandra _ have known rotation periods . to what extent
is the sample with periods representative of the entire onc population in terms of x - ray properties ?
if they are not representative , how does this group differ in other salient characteristics , such as accretion , and how might these differences affect our interpretation of the origin of x - rays in pms stars ?
in addition , while it is clear from previous analyses that x - ray emission from pms stars is not temporally static x - ray flaring is ubiquitous in the onc it is not yet clear whether or to what extent x - ray variability may be affecting our ability to measure reliable x - ray luminosities . might x - ray flaring be scrambling the signal of an underlying rotation / x - ray relationship ?
that some stars flare while others do not is interesting in its own right : do the flaring characteristics of stars with known rotation periods represent those of all stars ? again , how might differences here affect our interpretation of the origin of pms x - rays ? motivated by these questions , we have re - analyzed all archival _
chandra_/acis observations of the onc that include stars with known rotation periods .
our aim is to derive x - ray luminosities for as large a sample of known rotators as possible , employing a consistent analysis scheme throughout , including filtering of flares in the hopes of minimizing the effects of x - ray variability . in 2
we describe the data used and our processing / analysis procedures .
we then present our basic results in 3 , first focusing on the x - ray nature of the rotator sample as compared to the entire onc population .
we show that stars with known rotation periods are significantly more x - ray luminous , and more likely to be x - ray variable , than stars for which rotation periods have not been measured .
we then explore the relationship between x - rays and rotation .
we find that most stars with known rotation periods appear to be in the super - saturated regime , having @xmath11 , with a statistically significant correlation in which faster rotators have lower x - ray luminosities .
but we also find that stars without rotation periods being less x - ray luminous on average show a range of @xmath12 comparable to that observed on the main sequence .
these stars may represent the beginnings of the linear regime of the rotation / x - ray relationship .
finally , we explore the relationship between x - rays and accretion .
we find that while stars with spectroscopic signatures of accretion show harder x - ray spectra than non - accretors , they also show lower x - ray luminosities and no enhancement of x - ray flaring .
we discuss the implications of our findings in 4 , where we ( a ) emphasize that current rotation - period measurements in the onc have not probed the full range of underlying stellar x - ray properties , ( b ) suggest that a main - sequence type relationship between x - rays and rotation may in fact be present in the onc , and ( c ) argue that the data imply a chromospheric not accretion origin for x - rays from pms stars .
we summarize our conclusions in 5 .
our primary goal is to study the relationship of stellar x - rays to stellar rotation among a large sample of pms stars in the onc .
key parameters in our analysis are the stellar rotation period and the ratio of the x - ray luminosity , @xmath2 , to the bolometric luminosity , @xmath13 .
we restrict our analysis to observations with acis because its energy resolution allows @xmath2 to be determined from fits to the x - ray spectral energy distribution .
we also consider only reasonably long observations ( @xmath14 ksec ) so that we can attempt to derive quiescent x - ray luminosities by filtering out flaring events .
thus our study sample comprises onc stars that : ( 1 ) have known rotation periods , ( 2 ) have derived bolometric luminosities , and ( 3 ) have been observed by the _ chandra _ acis instrument . where possible
we would also like to study the relationship of x - rays to accretion , so we include such measurements where available . here
we describe the data from the literature that we compile to form our study sample ( [ literature ] ) .
we also describe the data from the _ chandra _ archive that we use ( [ chandra ] ) as well as the procedures employed in their reduction ( [ reduction ] ) and analysis ( [ analysis ] ) .
we close with a brief discussion of our assessment of the quality and reliability of the x - ray measurements ( [ quality ] ) .
rotation period measurements are available from the optical studies of @xcite and @xcite for 431 pms stars in the onc which were also included in the optical photometric / spectroscopic study of @xcite .
the latter study provides bolometric luminosities and other basic stellar parameters ( i.e. masses , effective temperatures , extinctions , etc . ) for most ( 358 ) of these stars .
in addition , the study of @xcite provides spectroscopic measures of accretion in the form of equivalent widths . in table [ table - sample ] we summarize our study sample , comprising 220 unique stars with rotation periods that we detect in the _ chandra _ observations described below .
we include relevant stellar properties taken from the sources above .
there are three onc observations in the _ chandra _ archive relevant to this study , two obtained by garmire ( obs . i d s 18 and 1522 ) and one by tsujimoto ( obs .
i d 634 ) .
the garmire observations are described by @xcite and include a 45.3 ksec exposure obtained on 1999 oct 1213 and a 37.5 ksec exposure obtained on 2000 apr 12 , both centered on the trapezium .
the tsujimoto observation , described in @xcite , is a single 89.2 ksec exposure centered on the omc-2/3 region ( just north of the trapezium ) obtained on 2000 jan 12 . in all three exposures ,
the four acis - i chips were operational with a total field of view of @xmath15 arcmin .
in addition , all three exposures had the acis - s2 chip in operation , which is separated from acis - i by 2.7 arcmin and has a field of view of 8.3 arcmin .
finally , the second garmire exposure included the acis - s3 chip , again with a field of view of 8.3 arcmin .
we include the acis - s data here for completeness , but note that this results in only a few additional sources due to the highly degraded point spread function ( psf ) of the instrument at large off - axis angles .
the acis instrument measures photon arrival times , positions , and energies ( 0.58 kev ) , so that for each detected source an x - ray light curve and spectral energy distribution can be constructed .
we reprocessed all three exposures in the same manner , starting from the _
evt1 event files , using the standard ciao procedure process_acis_events and updated calibration files obtained from the _ chandra _ x - ray center ( cxc ) in sept 2002 .
photon events were filtered according to their grade and status flags , and the images destreaked , following the standard cxc science threads .
we also manually updated the astrometric header keywords based on the latest astrometric calibration available from the cxc .
the resulting event files ( _ evt2 files ) were then searched for point sources using the ciao task celldetect .
the task uses a spatially variable psf , and we kept only those sources with a signal - to - noise ratio ( snr ) of 5 or greater .
we set the celldetect task to return source ellipses with a size of 99% encircled energy , and defined a background annulus whose inner and outer semi - major axes were , respectively , 1.5 and 1.7 times larger than the source ellipse .
to make the photon extraction computationally feasible , at each source position we then extracted a sub - region just larger than the background ellipse , using a set of idl scripts developed by us .
thus for each of the three exposures , the result of the reduction step is a set of event files , one for each of the detected sources . with a set of event files corresponding to each source detected with snr @xmath16 , we next applied an automated time - filtering of each source light curve in order to remove flare events prior to modeling the x - ray spectral energy distribution ( sed ) to derive x - ray luminosities .
the aim of this procedure is to determine a quiescent @xmath2 for each source .
based on the documented sensitivity limits of acis , in all that follows we use only x - ray photons with energies in the range 0.58 kev .
the time - filtering of the light curves was implemented in idl using procedures developed by us . for each source ,
the process involves the following steps ( see example in fig .
[ fig - lcfilt ] ) : ( 1 ) construct source and background light curves using the ciao lightcurve script with a binning interval of 2 ksec ; ( 2 ) subtract background light curve from source light curve ; ( 3 ) exclude bins that are @xmath17 brighter than the median , which is computed from the lowest 15% of the bins ; ( 4 ) re - determine the median and again exclude deviant bins , iterating until no more bins are excluded ; and ( 5 ) output a new event file that includes only the time intervals of the surviving bins . with time - filtered event files in hand for each of the detected sources
, we determined the @xmath2 of each source via a standard spectral analysis using sherpa . for each source in each of the three exposures , the position - dependent auxiliary response file ( arf ) and redistribution matrix file ( rmf )
were computed with the ciao psextract command and a model spectrum was fit .
the model used was a two - component thin thermal plasma with absorption by an intervening column of hydrogen .
the free parameters of the model are the absorbing hydrogen column density ( @xmath18 ) , the temperatures of the two plasma components ( @xmath19 , @xmath20 ) , the metallicity ( @xmath21 ) , and a normalization ( scaling ) coefficient for each plasma component .
a @xmath22 minimization procedure was used to fit each source s sed for these parameters , iterating until convergence was achieved . given the large number of free parameters , there is no guarantee that the best fit adopted is truly a global best fit or even that there is only one possible global best fit .
thus we emphasize that our goal in the spectral fitting is not the values of the model parameters themselves ; we simply seek a reasonably good fit from which we can determine the x - ray luminosity of the source .
the fit can thus be thought of as a ( possibly over - determined ) spline fit to the x - ray sed of the source , which we then integrate to measure the source flux , @xmath23 .
adopting a distance of 470 pc to the onc we convert the measured @xmath23 values into @xmath2 . in principle
, we can correct each @xmath2 for intervening absorption using the value of @xmath18 determined from the spectral fit .
however , @xcite have demonstrated that the @xmath18 values determined from spectral fitting do not correspond very well to @xmath24 values determined from optical photometry / spectroscopy .
we thus follow @xcite and choose not to correct the measured @xmath2 for absorption . with @xmath2 values determined for each source from each of the three _ chandra _ exposures ,
we match the sources with known rotation periods by searching for a positional match within the error ellipses of the detected x - ray sources .
we find 220 stars with rotation periods in the _ chandra _ images . in cases where a given target is detected in more than one _
chandra _ exposure , we select for our subsequent analysis the _ lower _ value of @xmath2 , assuming that the source changed its intrinsic @xmath2 between observations , and that the lower value represents the best estimate of the quiescent @xmath2 .
the x - ray properties of these 220 sources ( table [ table - sample ] ) are summarized in table [ table - lx ] , which includes all @xmath2 measurements of each source ( as many as three different measurements because there are three separate exposures ) .
in addition , table [ table - lx ] provides the @xmath2 measurements from @xcite for comparison .
those authors detected 253 stars with rotation periods , and here we re - detect 190 of them , presumably due to our higher snr criterion ( @xmath16 ) for source detection ( see above ) .
the 30 stars with rotation periods detected by us and not by @xcite ( see table [ table - lx ] ) derive from the @xcite exposure .
table [ table - lx ] also includes a descriptor for the variability of each source s light curve .
these are taken from @xcite when the source was included in that study ; otherwise , the descriptor is assigned by us following the procedure of @xcite .
a designation of ` const ' indicates that the light curve shows an approximately constant flux with time ; ` ltvar ' indicates statistically significant variability that occurs slowly in time , resulting in a different mean flux level in the different acis exposures ; ` flare ' indicates a statistically significant variation on short timescales ; and ` posfl ' indicates a flare - like variation of marginal significance .
finally , table [ table - lx ] provides a subjective quality flag for each @xmath2 determination , which we now discuss . since our re - reduction of the archival _ chandra _ data used updated calibrations , and
because our analysis procedures included time - filtering of flares that other authors have not done , in this section we assess the reliability of our reductions .
we begin by comparing the @xmath2 values derived by us to those derived by other authors for the same sources .
we then discuss some specific cases in detail in order to illustrate the vagaries inherent to this type of analysis . to start , we visually inspected the sherpa fit of each source and subjectively flagged those sources whose @xmath2 values we deemed unreliable due either to an observed spectrum with few counts or to an otherwise poor fit .
the result of this procedure is 154 sources whose spectra and corresponding spectral fits we felt were subjectively reasonable .
we restrict all subsequent discussion to these 154 sources , which are indicated in table [ table - lx ] by a quality flag of ` 1 ' . in fig .
[ fig - lxcomp ] we compare the @xmath2 values obtained by us to those obtained by @xcite for the common sources .
we basically find good agreement between the two sets of measurements .
a gaussian fit to the differences between the two measurements results in a standard deviation of @xmath25 dex , an offset of 0.15 dex ( our measurements being systematically larger ) , and a small number of outliers .
approximately 0.04 dex of the systematic offset can be accounted for by the fact that we assume a distance to the onc of 470 pc while @xcite assume a slightly lower value of 450 pc .
the remaining difference of 0.1 dex remains unaccounted for , but is not surprising given small differences in the calibrations used in our data reprocessing . on the whole , then , we can report reproducibility of the derived @xmath2 to a level of @xmath26 dex , despite differences in calibration , our time - filtering of flares from the light curves , and so on . nonetheless , a few stars have very different @xmath2 measurements from the two analyses ( up to about 1 dex ) . as an example , we consider star 116 , which is the most discrepant between our measurements and that reported by @xcite . from the two garmire exposures we measure @xmath2 values for this source of @xmath27 erg / s and @xmath28 erg / s , which encouragingly are similar to one another , but are very different from the @xcite value of @xmath29 erg / s ( see table [ table - lx ] ) .
this is a remarkable difference considering that these values derive from the same photons .
close inspection of our sherpa fits to the two observations of this source ( fig .
[ fig - star116-g1 ] ) do not indicate any obvious problems .
perhaps the discrepancy is the result of our flare filtering procedure .
however , the light curve of this source does not include any strong flares and so was not heavily filtered . in any case , we performed the sherpa analysis once again but on the pre - filtered data from the first garmire exposure .
as expected , the resulting @xmath2 of @xmath30 erg / s differs only slightly from the value we report in table [ table - lx ] , and the model fit again does not present any obvious problems ( fig .
[ fig - star116-test ] ) . recalling that the @xcite analysis typically used single - component fits to the spectra as compared to our two - component fits
, we attempted to reproduce their value by again running the sherpa analysis on the pre - filtered data but this time using only one thermal plasma component to the model fit .
the value of @xmath2 that we derive here ( @xmath28 erg / s ) still does not resolve the discrepancy , and may in fact be a low measure as the model fit in this case underestimates the flux in the two highest energy bins that are not upper limits ( fig .
[ fig - star116-test2 ] ) .
thus in this example case , and in the other discrepant cases seen in fig .
[ fig - lxcomp ] , we are simply unable to determine the cause of the discrepancy .
we provide this exercise as a cautionary lesson about the limits inherent in this type of analysis , but take comfort in the fact that for the majority of the sources used in our analysis the agreement between our values and those derived by @xcite is in fact very good .
the x - ray luminosities for each source in table [ table - sample ] resulting from our analysis are given in table [ table - lx ] , representing 220 stars with known rotation periods that are included in the optical database of @xcite . in this section
we report the results for the 154 sources having a quality flag of ` 1 ' .
we remind the reader that our values of @xmath2 are broadband luminosities over the energy range 0.5 kev to 8 kev , are not corrected for absorption , and do not include photon events that occur during a flare ( see [ analysis ] ) .
@xcite report @xmath2 measurements for an additional 63 stars with rotation periods detected at lower signal - to - noise ; where appropriate we include these measurements in our analysis and discussion , but in all cases we maintain a distinction between this larger sample and the subset which we believe to be of highest quality .
we begin by presenting the basic x - ray properties of these sources , emphasizing two biases that appear to be inherent to pms stars having measurable rotation periods ( [ basic ] ) , namely , a tendency toward higher x - ray luminosities ( [ lx ] ) and toward higher levels of x - ray variability ( [ var ] ) . with these biases in mind ,
we next examine the x - ray data vis - a - vis rotation ( [ rotation ] ) and accretion ( [ accretion ] ) for clues into the possible mechanisms for x - ray production in these stars . in this section
we discuss the basic x - ray properties luminosity and variability of stars with known rotation periods . by comparing these properties to those of other stars detected in the _ chandra _ observations , we find two biases astrophysical in origin in the rotation - period sample . stars with measured rotation periods are : ( 1 ) more x - ray luminous both absolutely ( i.e. @xmath2 ) and relative to the stellar bolometric luminosity ( i.e. @xmath12 ) ; and ( 2 ) more likely to be x - ray variable than are stars in the overall pms population of the onc . these results are highly statistically significant .
we emphasize that these biases are not due to observational bias ( e.g. , optical magnitude bias ) in the rotation - period sample , and are therefore likely to have a physical basis as we discuss in [ disc - rotation ] . here
we present the evidence for these two biases in turn .
we find that onc stars with known rotation periods are significantly biased to high x - ray luminosities . in fig .
[ fig - lxbias ] we plot both the distribution of @xmath3 for our study sample ( table [ table - lx ] ; hatched histogram ) as well as the larger sample of stars with rotation periods detected by @xcite ( dashed histogram ) . for comparison ,
the solid histogram shows the distribution of @xmath3 for all stars reported by @xcite included in the optical survey of @xcite . to demonstrate that the bias to high @xmath2 among stars with rotation periods is not due to optical bias in the rotation - period studies , we include here only those stars detected by @xcite having optical magnitudes bright enough ( @xmath31 ) to have been included in the optical samples studied for rotation periods @xcite .
we further restrict this comparison sample to only stars with masses @xmath32 , as this represents the range of stellar masses among stars with rotation period measurements .
while the stars with known rotation periods ( dashed histogram ) exhibit a range of @xmath2 , this range is @xmath33 dex smaller than that spanned by the underlying onc population ( solid histogram ) .
moreover , the @xmath2 distribution of these stars is skewed with respect to the overall distribution , such that stars with rotation periods exhibit higher average @xmath2 . to show this more clearly
, the distribution of @xmath2 for stars _ without _ rotation periods ( i.e. the difference between the solid and dashed histograms ) is shown also ( dot - dashed histogram ) .
a two - sided k - s test indicates that the probability of the @xmath2 distributions for stars with and without rotation periods ( dashed and dot - dashed histograms ) being drawn from the same parent population is @xmath34 .
in addition , a student s @xmath35 test gives a probability of only @xmath36 that the means of these two distributions ( @xmath37 erg / s for stars with rotation periods and @xmath38 erg / s for stars without ) are the same .
a similar result is obtained when we consider @xmath0 instead of @xmath2 ( fig . [ fig - lxlbolbias ] ) . here
, a two - sided k - s test gives a probability of @xmath39 that the @xmath0 distributions for stars with and without rotation periods are drawn from the same parent population .
and a student s @xmath35 test gives a probability of @xmath40 that the means of these two distributions ( @xmath41 for stars with rotation periods and @xmath42 for stars without ) are the same .
a similar bias manifests itself with respect to x - ray variability of the sources .
the subset of stars in our sample whose x - ray light curves are variable ( ` flare ' , ` posfl ' , or ` ltvar ' in table [ table - lx ] ) comprise 82% @xmath43 3% of our study sample ( uncertainties determined from the binomial distribution ) .
similarly , 70% @xmath43 3% of stars with rotation periods in the larger sample of @xcite show variability . in comparison , a smaller fraction , 57% @xmath43 2% , of onc stars in the @xcite study that lack rotation periods
show such variability . for the entire sample of stars with rotation periods , this difference in x - ray variability
is statistically significant .
a @xmath22 test gives a probability of 0.001 that stars with and without rotation periods have equal occurrences of variability .
for our high - quality sample , where the signal - to - noise is higher and variability in the light curves is therefore better determined , a @xmath22 test gives a probability of @xmath44 that the occurrence of variability is the same as that found among stars without rotation periods .
there thus appears to be significant evidence for an enhancement of x - ray variability among stars in the onc with rotation periods , particularly when we restrict our analysis to those stars with the highest quality x - ray light curves .
x - ray emission on the main sequence among stars with @xmath45 m@xmath46 is believed to be driven by stellar rotation , and this results in a clear , observable correlation between stellar rotation and x - ray luminosity .
the relationship between x - ray luminosity and stellar rotation period for our study sample is shown in fig .
[ fig - lxvsrot ] , where we plot @xmath47 vs. @xmath48 . for ease of comparison ,
the vertical scale is set to the full range of @xmath47 observed on the main sequence . as noted above and in the previous studies of @xcite and @xcite
, these stars show a mean @xmath47 near the main sequence saturation value of @xmath49 , though somewhat lower ( mean @xmath41 for all stars with rotation periods ) .
taken at face value , these data present no clear evidence for an x - ray / rotation relationship of the sort seen on the main sequence . at a more detailed level , these data provide possible evidence for these stars being in the super - saturated regime of the rotation / x - ray relationship .
in addition to having a mean @xmath12 below the saturation value , the data in fig . [ fig - lxvsrot ] also show a weak , but statistically significant , trend of increasing @xmath0 with increasing rotation period , as might be expected for stars in the super - saturated regime . among all stars with rotation periods ,
a spearman s @xmath50 rank - correlation test gives a probability of @xmath51 that @xmath52 is uncorrelated with @xmath12 .
the same trend is present among the smaller set of stars detected in this study , though only at 95% significance . to effect a better comparison with super - saturation on the main sequence , we transform the abscissa from @xmath52 to rossby number , @xmath53 , defined as the ratio between @xmath52 and the convective turnover timescale , @xmath54 ,
is typically determined from stellar interiors models for stars of the appropriate mass and age . as discussed by @xcite , at the young age of the onc
the value of @xmath54 is roughly constant for these fully convective low - mass stars .
we thus convert @xmath52 to @xmath53 by scaling the former by a constant value of @xmath55 days @xcite .
] , which is typically used to show the x - ray / rotation relationship on the main sequence .
this is shown in fig .
[ fig - lxvsrossby ] , where the solid line represents the main - sequence relationship as determined by @xcite , and where the stars in our sample now appear explicitly in the super - saturated regime .
[ fig - lxvsrossby ] also shows the @xmath47 for the remainder of the onc sample from @xcite with @xmath32 ( crosses plotted arbitrarily at @xmath56 ; these are the same stars as in the dot - dashed histogram in fig .
[ fig - lxlbolbias ] ) .
as we have seen ( [ lx ] , figs .
[ fig - lxbias ] and [ fig - lxlbolbias ] ) , these stars are on average less x - ray luminous than are stars with known rotation periods .
might there also be differences on average in their rotational properties ? for 40 of these stars lacking optical rotation periods ,
@xmath57 measurements are available from the study of @xcite , allowing us to infer their ( projected ) rotational characteristics . in fig .
[ fig - vsini ] we show the @xmath2 distribution for these stars segregated into two groups , fast ( 11 stars ) and slow rotators ( 29 stars ) , defined on the basis of whether @xcite report a @xmath57 measurement or a @xmath57 upper limit ( i.e. whether the spectral lines are broadened beyond the instrumental resolution or not )
. the slow rotators indeed appear to be skewed to lower @xmath2 , and both a two - sided k - s test and a student s @xmath35 test confirm this at the 99% confidence level .
the difference between slow and rapid rotators is not statistically significant when we consider @xmath12 instead of @xmath2 .
a similar test is possible among stars with @xmath58 measurements that _ do _ have rotation periods ( 58 fast and 62 slow rotators ) .
the @xmath2 distributions of these two groups are statistically indistinguishable .
apparently , the difference in @xmath2 between fast and slow rotators is only present among stars lacking optical rotation periods .
thus , while there is not a one - to - one correlation between @xmath2 and @xmath57 for stars without optical rotation periods , there is a marginally significant tendency for the x - ray faint stars in this group to also have slower rotation speeds .
this is in the opposite sense to what we find above for stars that do have optical rotation periods , in which the x - ray luminosity _ increases _ with slower rotation similar to super - saturated stars on the main sequence ( cf .
fig . 9 in @xcite ) , albeit with a large scatter .
accretion is another mechanism possibly related to x - ray production in pms stars , and indeed accretion appears to manifest itself strongly in the x - ray properties of the stars in our study .
we use the strength of emission in the line as measured by @xcite to determine which stars are actively accreting : following @xcite we take stars with equivalent widths ( ew ) of @xmath59 ( i.e. in emission ) to be those actively accreting , while those with ew @xmath60 ( i.e. in absorption ) to be non - accreting .
ew measurements are available for 117 stars in our sample and for 199 stars among all stars with rotation periods . in light of the biases inherent to the rotation period sample noted in
[ basic ] , where appropriate we also explore accretion signatures in the full sample of onc stars from the study of @xcite .
we find that stars with active accretion signatures in , while no more likely to show x - ray flares than non - accreting stars , are systematically less x - ray luminous and exhibit systematically harder x - ray spectra .
we discuss in turn the relationship between accretion and x - ray flaring , x - ray luminosity , and x - ray hardness .
we begin by noting that spectroscopic signatures of active accretion are relatively rare among the stars in our sample . among the 117 stars from this study that have measurements ,
only 10 stars show clearly in emission ( i.e. ew @xmath59 ) , whereas 66 stars show clearly in absorption ( i.e. ew @xmath60 ) . among those few stars that do show evidence for active accretion , all 10 of them exhibit x - ray flaring in the _ chandra _ data ( ` flare ' or ` posfl ' in table [ table - lx ] ) . among the non - accreting stars , 70% ( 46/66 stars ) show such evidence for x - ray flaring . because of the small number of accreting sources in this sample , this difference is not statistically significant .
similarly , among the larger sample of all stars with rotation periods only 28/199 stars show clearly in emission , whereas 77 stars show in absorption . among the 28 accreting stars ,
15 ( 54% ) show evidence for x - ray flaring , while among the non - accreting stars 47 stars ( 61% ) do .
this small difference is not statistically significant . considering the entire onc sample included in the study of @xcite ,
there are 254 stars for which @xcite report a ew of either @xmath59 ( 126 stars ) or @xmath60 ( 128 stars ) . in this larger sample ,
41% of the accreting stars show x - ray flaring , and 48% of the non - accreting stars do , again indicating no relationship between accretion and x - ray flaring .
we thus find that while stars with optical rotation periods are predominantly non - accreting ( see also @xcite ) , x - ray flaring is nonetheless ubiquitous among them ( [ var ] ) , and the presence of active accretion does not significantly enhance this x - ray flaring . among the stars with measured rotation periods ,
we find a hint that actively accreting stars have lower x - ray luminosities than their non - accreting counterparts .
as above , there are only 28 stars with rotation periods that show clear signs of active accretion and 77 stars that clearly do not . comparing the @xmath2 distributions of these two subsets , a student s @xmath35 test reveals different means with accretors being less luminous at 98% confidence .
however , within the full onc sample we find that this difference in @xmath2 between accretors and non - accretors is highly statistically significant .
of the 529 stars from @xcite in the _ acis _ field , @xcite detect 525 stars with @xmath32 . of these ,
256 have ew ( ) @xmath59 ( 126 detected in x - rays , 0 undetected ) or ew ( ) @xmath60 ( 128 detected in x - rays , 2 undetected ) . here
we ignore the two undetected stars . as fig .
[ lx - acc]a shows , the @xmath2 distributions of accretors and non - accretors are clearly different ; a two - sided k - s test reveals that the probability that the two are drawn from the same parent distribution is @xmath61 .
as demonstrated by @xcite and @xcite , @xmath2 correlates strongly with stellar mass .
thus , the differences in @xmath2 among accretors and non - accretors might be the result of a correlation between accretion and stellar mass .
[ lx - acc]b shows the @xmath2 distributions for accretors and non - accretors as a function of mass ( stellar masses taken from @xcite ) .
the center of each box markes the position of the median @xmath2 in that mass bin .
if the indented regions around the medians ( notches " ) of two boxes do not overlap , the medians are different with @xmath62 95% confidence ( see @xcite for an explanation of box plots ) .
we see that for stars below @xmath33 m@xmath46 , those with spectroscopic accretion indicators have significantly lower @xmath2 than stars that do not have spectroscopic accretion indicators .
the number of objects in the higher mass bins , particularly those showing active accretion in , is sufficiently small that the uncertainties on the boxes in fig .
[ lx - acc]b are large and any differences between accretors and non - accretors may be difficult to detect .
in addition to x - ray luminosity , the _ chandra / acis _ data allow us to compare accretors and non - accretors in terms of x - ray spectral properties .
[ hr - acc]a compares the histograms of hardness ratios [ hr @xmath63 for accretors and non - accretors , where @xmath64 is the x - ray luminosity from 0.5 to 2 kev , and @xmath65 is the x - ray luminosity from 2 to 8 kev . as above , we include in our analysis all onc stars from the study of @xcite with @xmath66 m@xmath46 .
we find that accretors exhibit systematically harder x - ray spectra than non - accretors , and the likelihood of both samples being drawn from the same parent distribution is @xmath67 .
[ hr - acc]b shows the mass dependence of the hr .
similar to fig .
[ lx - acc]b , a difference between accretors and non - accretors is clear for stars with masses below @xmath33 m@xmath46 .
from our analysis of all archival _ chandra / acis _ observations of a large sample of pms stars in the onc , we have identified important biases in the basic x - ray characteristics ( luminosity and variability ) of stars with optically determined rotation periods as compared to the overall population of pms stars detected by _
chandra_. in addition , we have explored possible relationships between the x - rays observed from these stars and the two physical mechanisms most likely responsible for their production : rotation and accretion . in this section
we explore in greater depth the implications of the findings presented in [ results ] toward the goal of further elucidating the origin of x - rays in pms stars .
we structure this discussion again around the two central physical mechanisms of rotation and accretion .
we will argue that the data hint at the presence of an underlying rotation / x - ray relationship qualitatively similar to that observed on the main sequence , and we will show that the observed differences in x - ray characteristics between accretors and non - accretors are in fact consistent with a picture in which all stars have intrinsically similar x - ray emission properties . we therefore posit that rotation and not accretion is primarily responsible for the production of x - rays in pms stars at @xmath8 myr . in seeking to find a rotation / x - ray relationship among pms stars analogous to that observed on the main sequence , it is logical to focus on the x - ray properties of pms stars with known rotation periods .
unfortunately , the full rotation / x - ray relationship , if it exists among pms stars in the onc , might not be discernable from those stars with optically determined rotation periods alone .
as we have seen , these stars are significantly biased to higher values of @xmath2 ( and @xmath12 ) than are stars without rotation periods .
these stars may therefore only allow us to probe the super - saturated regime of any underlying rotation / x - ray relationship .
why are pms stars with optically determined rotation periods biased in their basic x - ray characteristics ?
it appears that this bias results from the fact that rotation periods can only be measured among stars with spots that are sufficiently large and long - lived to produce stable periodic signals in the optical . to show this , in fig .
[ lx - spots]a we plot the amplitude of optical variability , @xmath68 , as reported by @xcite for pms stars in the onc with rotation periods , against these stars x - ray luminosities as determined in this study and in the study of @xcite .
the two quantities are highly correlated .
whether we consider all stars with rotation periods , or only those detected in this study ( filled circles in fig .
[ lx - spots ] ) , a spearman s rank - correlation analysis yields a probability of @xmath69 that @xmath68 and @xmath2 are uncorrelated
. the same result is obtained when we consider @xmath12 instead of @xmath2 ( fig .
[ lx - spots]b ) . in this case
, we find a correlation at marginal confidence ( 99% ) when we consider only the stars from this study , but a probability of @xmath70 that the two quantities are uncorrelated when we include all stars with rotation periods .
the implication is that we do not observe stars with rotation periods at very low @xmath2 because the amplitude of photometric variability in the optical becomes diminishingly small , ultimately smaller than the minimum signal detectable ( @xmath71 mag ) by existing rotation - period studies of the onc @xcite . in light of the fact that stars with rotation periods
have high x - ray luminosities , it is perhaps not surprising that these stars appear to be in the super - saturated regime .
but if these stars are indeed super - saturated as fig .
[ fig - lxvsrossby ] implies , then the optical variability data would seem to imply a qualitatively different picture for the surfaces of super - saturated stars than that commonly assumed .
the mental image often invoked in the context of saturation is that of a star whose surface has become completely threaded by magnetic flux tubes , resulting in spot coverage fractions approaching unity . yet in fig .
[ lx - spots ] there are stars at both low and high @xmath12 that show relatively small amplitudes of optical variability , suggesting that spot coverage among many of these super - saturated " stars is relatively light . on the other hand , fig .
[ lx - spots ] may be telling us that these stars do indeed have spots covering large fractions of their surfaces , but that we are seeing changes in the magnetic topologies of these stars as a function of @xmath12 .
for example , stars at lower @xmath12 may represent stars with relatively disorganized surface fields that produce relatively small spots more - or - less uniformly distributed on the stellar surface .
such small , uniformly distributed spots would produce only low - amplitude variability in the optical even if they cover a large fraction of the stellar surface .
in contrast , stars with larger @xmath12 could represent cases where the magnetic field has become more coherently organized into relatively large spots that are distributed more asymmetrically on the stellar surface , thereby giving rise to larger photometric variability in the optical . that not all stars with large @xmath12 have correspondingly large @xmath68 is perhaps simply due to geometrical effects ( varying spot sizes / temperatures , spot latitudes , inclination angles , etc . ) , or it may suggest that strong magnetic fields do not instantaneously arrange into organized configurations .
this interpretation is similar to that proposed by @xcite , who argues that stars in the super - saturated regime are cases in which the stellar magnetic field has not yet become sufficiently organized to couple the stellar interior to the surface , and therefore the star s rotation is not effectively braked . @xcite
further argues that as the stellar magnetic field becomes more organized and achieves maximum strength , it becomes more deeply rooted , the x - ray luminosity also reaches maximum strength ( saturation ) , and magnetic braking begins to affect the entire star . in this way , @xcite offers a possible explanation for the positive correlation observed between @xmath52 and @xmath12 among stars in the super - saturated regime .
these are speculative ideas to be sure ; our aim here is to provide additional observational fodder to the question of what super - saturation is really telling us about the magnetic nature of pms stars . at any rate ,
if we accept the inference that stars in the onc with rotation periods do represent the super - saturated regime of the rotation / x - ray relationship , then the question arises whether there is evidence for an unseen linear regime in the rotation / x - ray relationship .
[ fig - lxvsrossby ] tells us that there are indeed stars with sufficiently low @xmath0 , but do these stars also rotate more slowly ? while the available @xmath58 data do not show a one - to - one relationship between @xmath58 and @xmath2 , we do find evidence that slower rotators do indeed have lower @xmath2 ( [ rotation ] ) , hinting at behavior qualitatively consistent with the linear regime of the rotation / x - ray relationship .
thus , a picture begins to emerge from the data in which x - ray luminosity does appear to be related to stellar rotation among pms stars in the onc .
stars with rotation periods , biased as they are in @xmath2 , may represent the super - saturated and saturated regimes , and some stars lacking rotation periods may represent the saturated and ( at least part of ) the linear regime , implying a population of very slow rotators among these stars . an alternative to the slow - rotator explanation for the lower @xmath2 of stars without rotation periods
is that stars without rotation periods are predominantly active accretors , and that it is accretion that is acting to suppress the @xmath2 of these stars ( see [ acc - lum ] ) .
indeed , among the sample of stars from @xcite that lack rotation periods , those with spectroscopic signatures of active accretion ( i.e. ew ( ) @xmath72 ) outnumber those without such signatures by 2:1 . to examine this possibility more fully ,
we have compared the hardness ratios ( hrs ) of stars with and without rotation periods , since hr is also correlated with accretion ( accretors produce harder hrs ; see [ acc - hardness ] ) .
we find that the hrs of stars without rotation periods are marginally harder than those with rotation periods ; a k - s test yields a probability of 1% that the distributions of hrs for the two groups are the same .
compared to the result in [ acc - hardness]where we found a highly statistically significant difference in hr for accretors vs. non - accretors this suggests that , for the particular mix of stellar masses and accretion properties in the non@xmath73 sample , accretion is only weakly related to the lower average @xmath2 of these stars .
the significance of the effect is , nonetheless , comparable to the @xmath58 effect described above .
discerning whether , or to what extent , the lower average @xmath2 of stars lacking rotation periods is due to accretion or slower rotation remains an open observational question .
unfortunately , the existing @xmath58 study of @xcite did not have sufficiently high spectral resolution to place stringent lower limits on the rotation rates of these stars .
it would thus be valuable to have high - resolution @xmath58 measurements targeting stars with very low @xmath2 and lacking @xmath52 in order to better constrain the slow extremes of rotation among stars that may represent the saturated and linear regimes ( @xmath74 days ) of the rotation / x - ray relationship .
finally , we call attention to the fact that stars with rotation periods , despite evincing stable optical photometric variability with low levels of stochasticity ( else their rotation periods would be difficult to measure ) , nonetheless show elevated levels of variability in x - rays ( [ var ] )
. this may suggest that the mechanism(s ) responsible for x - ray variability are decoupled from the mechanism(s ) often attributed to stochastic optical variability in pms stars ( i.e. accretion ) , as we now discuss .
it is now generally accepted that most , if not all , pms stars undergo a phase of active accretion whereby circumstellar material , perhaps channeled by stellar magnetic field lines , is deposited onto the stellar surface .
models of this accretion process @xcite have had some success in explaining the continuum excesses often observed in the uv among pms stars as being due to the energetic shock that arises when accreted material impacts the stellar surface .
accretion is also typically implicated as the source of the stochastic , optical variability that is a defining characteristic of classical t tauri stars ( ctts ) @xcite .
it is appropriate to ask , therefore , whether x - rays from pms stars may also have their origins , at least partly , in accretion .
we have already seen that x - ray variability is ubiquitous among the pms stars in this study , despite the fact that the majority of these stars are weak - lined t tauri stars ( wtts ) , as they do not show spectroscopic indicators of active accretion ( [ acc - flaring ] ) .
but perhaps accretion acts nonetheless to noticeably affect the x - ray emission of these stars .
indeed , we have seen that accretors and non - accretors do differ both in their x - ray luminosities ( [ acc - lum ] ) and x - ray hardness ( [ acc - hardness ] ) . here
we investigate these differences in greater detail .
we begin by reviewing the evidence , both from this study and from others in the literature , for a difference in the x - ray luminosities between accretors and non - accretors .
we then present a simple model that explains these differences naturally in terms of enhanced x - ray _ absorption _ among stars with active accretion , due to the presence of magnetospheric accretion columns . among pms stars in a variety of star formation regions ,
there appears to be strong evidence for a difference in x - ray luminosity between accretors and non - accretors , in the sense that accretors tend to be underluminous in x - rays relative to non - accretors .
a summary of the situation with a re - analysis of _ rosat _ data is presented in @xcite for the onc , ngc 2264 , and chameleon i. similar results are found by @xcite in taurus - auriga . however , the most recent observations in orion present two different results .
@xcite find that that the difference in the median @xmath2 between accreting and non - accreting stars is about one order of magnitude in the 0.252 m@xmath46 range , in agreement with the earlier _ rosat _ findings .
these authors use the ew of the lines , as reported by @xcite , to distinguish accretors from non - accretors .
their study is based on a single exposure with _
chandra / hrc _ ( 30 by 30 ) centered on @xmath75ori c. optical observations catalog 696 cluster members in the field , 342 of which are detected in the hrc image .
of the 696 possible members , a subset ( 304 stars ) have ew ( ) @xmath59 ( 108 x - ray detected , 58 undetected ) or ew ( ) @xmath60 ( 54 x - ray detected , 84 undetected ) . as the hrc instrument does not provide spectral information , the authors assume a fixed plasma temperature for all sources and gas column density proportional to optical extinction in order to derive x - ray luminosities .
in contrast , @xcite find no difference in the distributions of ctts and wtts with respect to x - ray luminosity . here , the distinction between ctts and wtts is made in terms of k - band excess , which is taken to indicate the presence of an accretion disk .
their study is based on the same _ chandra / acis _ observations that we use in our own analysis . the acis image ( 17 by 17 )
is centered 22 west of @xmath75ori c. in that region there are 529 optically detected stars , 525 of which are detected in the acis exposure .
the discrepancy between the findings of @xcite and @xcite can be resolved by noting that while infrared indicators signal the presence of a disk , this does not necessarily signal the presence of active accretion : the presence of a disk is presumably a prerequisite for accretion to occur , but not necessarily vice - versa . indeed , using the same spectroscopic proxy for accretion as @xcite , our analysis above ( [ acc - lum ] ) confirms the findings of @xcite within the same acis observations used by @xcite .
we thus take the finding of a difference in x - ray luminosity between accretors and non - accretors , as shown in fig .
[ lx - acc ] , to be secure .
in addition , we have found evidence for a difference between accretors and non - accretors in terms of x - ray hardness ( fig .
[ hr - acc ] ) .
we now proceed to examine possible explanations for these differences .
pms stars undergoing active accretion show systematically lower x - ray luminosities and harder x - ray hardness ratios ( hr ) than their non - accreting counterparts .
this suggests that either : ( a ) the x - ray emission from accretors is intrinsically different in its spectral properties , namely , more concentrated to higher x - ray energies ( i.e. harder ) ; or ( b ) the x - ray emission from the accretors is intrinsically similar to that from non - accretors , but has been processed by circumstellar gas , preferentially attenuating x - rays at softer energies .
in the magnetospheric picture of accretion , ctts are encaged in funnels of inflowing gas @xcite with densities ranging from @xmath76 to @xmath77 @xmath78 @xcite .
these funnels may be 0.1 r@xmath46 thick , which implies that hydrogen column densities larger than @xmath79 @xmath80 are possible .
the exact amount of gas column will depend on the accretion rate and on the detailed geometry of the accretion flows but , as we show below , this amount of hydrogen column is potentially sufficient to both attenuate and harden the x - rays observed from ctts . to investigate this further , we first need to obtain the intrinsic ( corrected for ism absorption ) x - ray characteristics of the _ chandra _ sources .
the x - ray luminosities and hrs that we have so far used in our analysis have not been corrected for the attenuation and hardening caused by absorption due to interstellar gas . in some star formation regions , this is an important issue .
for example , @xcite have shown that in taurus the reddening toward ctts is significantly higher than toward wtts , which could produce systematic differences in @xmath2 and hr similar to what we have observed . in the @xcite data
there is no evidence for a systematic difference in extinction between accretors and non - accretors ; the extinction properties of both groups are the same to within 20% .
nonetheless , there may still be individual differences in extinction that could act to alter the medians in figs .
[ lx - acc ] and [ hr - acc ] . in order to correct for interstellar reddening
we have performed the following analysis .
we first calculate hr and @xmath2 values for a grid of hydrogen column densities and plasma temperatures ( fig .
[ model ] ) . to generate these models
, we used the xspec code @xcite , version 11.2 , assuming a uniform plasma with @xmath81 solar elemental abundances .
as in @xcite , continuum and line emission strengths were evaluated using the mekal code @xcite , and x - ray absorption was modeled using the cross sections of @xcite . for each star in the @xcite database , we take the hr and @xmath2 values reported by them and extinctions ( @xmath24 ) from @xcite .
we then use the relation @xmath82 to convert the observed extinctions into a measure of the hydrogen column density toward each star . from fig . [ model ]
, we obtain the ratio between the observed luminosity and the luminosity corrected for reddening .
for example , if a star is observed to have hr @xmath83 and @xmath84 , fig .
[ model ] tells us that the observed @xmath85 ( arbitrary units ) and that the intrinsic @xmath86 ( obtained by moving in constant @xmath87 to @xmath88 ) , implying that the x - ray luminosity has been extincted by a factor of @xmath89 and that the true hr is @xmath90 . in this way
we obtain corrected values of @xmath2 and hr for each star .
the results are shown in figs .
[ corr_boxes ] and [ corr_hard ] .
the temperature obtained by this procedure should be regarded as an `` effective '' plasma temperature , as individual fits suggest that in some cases multiple plasmas , each with a different temperature , are necessary to reproduce the observations .
the procedure also assumes that the plasma is in ionization equilibrium @xcite . after correcting for reddening ,
the differences in the histograms persist ( figs . [ corr_boxes]a and [ corr_hard]a ) , although when plotted as functions of mass ( figs .
[ corr_boxes]b and [ corr_hard]b ) the differences between the accretors and non - accretors become more subtle .
it is therefore legitimate to ask whether the differences in the histograms are real , considering the dependence of @xmath2 and hr on mass .
for example , the presence of proportionately more non - accretors than accretors at higher masses could potentially explain the differences in the histograms .
a two - way analysis of variance indicates that the @xmath2 averages of accretors and non - accretors , after eliminating the effect of the mass , have a probability of @xmath91 of being the same . for hr ,
the probability is @xmath92 . in other words
, there is a statistically significant difference between accretors and non - accretors , both in hr and in @xmath2 , even after controlling for the mass dependence .
interestingly , hr appears to increase ( albeit weakly ) with mass the analysis of variance indicates that the probability of all the means in the mass bins being the same is @xmath1perhaps implying that more massive stars have hotter chromospheres .
this is not due to the fact that higher mass stars have higher @xmath2 ; hr is scale - independent , so overall increases in @xmath2 do not affect it .
differences in hr values between ctts and wtts have been reported in the literature for taurus , lupus , chameleon , sco - cen , and the tw hya association @xcite .
all these are based on _ rosat _ data , for which two different hardness ratios are traditionally defined in the literature : hr1 @xmath93where @xmath94 is the count rate from 0.5 to 0.9 kev , @xmath95 is from 0.9 to 2 kev , and @xmath96 is from 0.1 to 0.4 kev and hr2 @xmath97 .
note that the two hard " _ rosat _ bands are equivalent to the soft " _ chandra _ band so the results from _ chandra _ and _ rosat _ are not directly comparable . in the _ rosat _ observations , and for these star formation regions ,
the wtts are as a group significantly softer than the ctts in the hr1 ratio , while the two populations have similar hr2 ratios .
our analysis shows that the difference reappears in the higher energy _
hr ratio , which samples energies up to 8 kev .
@xcite , finding no difference in emission temperatures between ctts and wtts in taurus , and considering different star - forming regions with different extinction characteristics , argue that this difference in hr is due to absorption in the circumstellar environs of the ctts ( circumstellar disks , remnant nebulae and envelopes , outflows , etc . ) . for the observations presented here , the differences in @xmath2 and hr between accretors and non - accretors are consistent with a picture in which ctts have intrinsically similar x - ray emission properties as wtts , with x - rays from the former being extincted by circumstellar gas in amounts consistent with that predicted for magnetospheric accretion columns .
the median hr ( corrected for absorption ) of the non - accretors in our sample is @xmath98 with @xmath99 . for the accretors ,
the value is @xmath100 with @xmath99 . assuming that the difference is due to gas absorption
, we can use fig . [ model ] to obtain the gas column density .
if the mean hr of the non - accretors in our sample ( @xmath98 ) represents the intrinsic hr of a t tauri star , this implies ( following the @xmath101 curve ) a plasma temperature of @xmath102 kev .
the curves are marked in dust extinction magnitudes , but in this excercise we are using them to correct for gas absorption only .
if we follow the line of constant @xmath87 to higher hardness ratios , we reach hr @xmath103 at @xmath104 , which implies @xmath105 @xmath80 . in this case
, the ratio in @xmath2 between accretors and non - accretors would be @xmath106 . given the width of the hr histograms , column densities as large as @xmath79 @xmath80 of gas may be necessary .
these produce ratios in @xmath2 as large as 0.8 dex , which is consistent with fig .
[ corr_boxes ] and with the results of @xcite . on the other hand , @xcite argue , on the basis of _ chandra _ x - ray spectroscopy , that the x - ray emission from tw hya is due to the accretion shock at the base of the accretion column , and not simply to attenuated wtts emission .
the differential emission meassure is quite unlike that of other active evolved stars ( even though it is not clear what one should expect for a pms star ) .
tw hya is a 10 myr old , 0.7 m@xmath46 pms star with @xmath107 erg / sec , and so it has a very average position in our @xmath2 vs. mass diagram .
this star poses a puzzle for the arguments presented here in favor of a common origin for x - ray emission in ctts and wtts .
its accretion rate has been reported as being 5100 @xmath108 m@xmath46/yr @xcite , and if the lower limit is right , one would expect essentially no gas attenuation .
in addition , coronal activity decreases with age , and so perhaps the observations of tw hya are not applicable to younger samples
. certainly , x - ray spectroscopic observations of young wtts and ctts are needed before this issue can be fully resolved .
we have re - analyzed all archival _ chandra / acis _ observations of pre main - sequence ( pms ) stars with optically determined rotation periods in the orion nebula cluster ( onc ) .
our aim is to investigate the relationship between x - rays and the physical mechanisms most likely related to their production in pms stars : rotation and accretion .
our analysis procedures include filtering of flare events in the x - ray data in an attempt to determine x - ray luminosities that are free of the stochasticity introduced by such events .
1 . stars with optically determined rotation periods are more x - ray luminous , and are more likely to be x - ray variable , than are stars without optical rotation periods .
we show that the bias to high @xmath2 is not due to a magnitude bias in optical rotation - period studies of the onc ; rather , it is due to the diminishingly small amplitude of optical variability among stars with smaller @xmath2 , precluding detection of their rotation periods .
stars with optically determined rotation periods have a mean @xmath12 near , but lower than , the saturation " value of @xmath109 , implying that these stars are in the saturated or super - saturated regimes of the x - ray / rotation relationship , consistent with their rossby numbers .
there is a marginally significant ( @xmath110 ) correlation between @xmath12 and @xmath73 , with the more rapidly rotating stars showing lower @xmath12 , as is seen among super - saturated stars on the main sequence .
3 . compared to these stars , stars _ without _ rotation periods show a larger range of @xmath12comparable , in fact , to that found among main sequence stars .
we consider the possibility that , among these , some stars may lie at the beginnings of the linear " regime of the x - ray / rotation relationship . using @xmath58 data from the literature we find that , among these stars lacking known rotation periods , slower rotators do indeed show lower x - ray luminosities than do rapid rotators .
this relationship is not one - to - one , however .
it is also possible that the lower @xmath2 among stars lacking rotation periods is instead due to the higher incidence of active accretion among these stars , a possibility for which we also find weak evidence .
the statistical significance of these two effects@xmath58 and accretion
are comparable . measurements of @xmath58 sensitive to very slow rotators ( @xmath111 km / s ) would be of great value in furthering our understanding of x - ray production at the slow extremes of pms rotation .
pms stars in the linear regime should have @xmath112 days , assuming a typical convective turnover timescale of @xmath10 days .
such long rotation periods have yet to be observed among pms stars .
stars in the onc with spectroscopic signatures of active accretion show significantly harder x - ray spectra and lower x - ray luminosities than their non - accreting counterparts .
these observations can be explained quantitatively by a model in which accretors and non - accretors have intrinsically similar x - ray emission properties , with the differences in @xmath2 and hardness ratio being due to absorption of soft x - rays by magnetospheric accretion columns . taken together , these findings hint that there in fact exists a rotation - activity relationship among pms stars in the onc , and suggest that rotation not accretion is the primary driver of x - ray emission in low - mass ( @xmath45 m@xmath46 ) pms stars at 1 myr .
indeed , our finding that stars with rotation periods show elevated levels of x - ray variability , despite showing little stochastic variability in the optical , further implies that x - ray variability has its origins in processes that are more or less independent of the processes responsible for stochastic variability in the optical ( i.e.accretion ) .
finally , our findings raise questions about the true physical meaning of saturation " in pms stars .
it is intriguing that stars with optically determined rotation periods all appear to lie in the super - saturated regime yet show diminishingly small amplitudes of optical variability at low @xmath2 .
it is possible that spots on the surfaces of these stars become non - existent below a certain @xmath2 threshold . on the other hand , we speculate that the low amplitude of optical variability may be due to magnetic topologies
in which the stellar surface is indeed largely covered by spots , but spots that are more - or - less randomly distributed over the stellar surface , thereby producing only very small photometric signals in the optical .
more organized magnetic topologies may be present in stars with higher @xmath2 , such that larger spots asymmetrically distributed on the stellar surface are possible . in this picture
, these latter stars might be those whose global fields have become sufficiently organized and deeply rooted so as to begin effecting magnetic braking of the stellar rotation , a picture similar to that recently put forward by @xcite .
we acknowledge funding under chandra award number ar2 - 3001x issued by the chandra x - ray observatory center , which is operated by the smithsonian astrophysical observatory on behalf of nasa under contract nas8 - 390073 .
we also gratefully acknowledge the useful comments of the anonymous referee .
rrrrrr 106 & 1.70 & 0.21 & @xmath1130.29 & 0.10 & 1.5 + 111 & 4.94 & 0.42 & @xmath1130.15 & 0.24 & 2.2 + 116 & 2.34 & 0.69 & 0.20 & 0.09 & 1.6 + 118 & 1.07 & 0.13 & @xmath1130.61 & @xmath1130.32 & 0.0 + 122 & 0.98 & 0.14 & @xmath1130.54 & @xmath1130.17 & 0.0 + 123 & 6.63 & 1.37 & 0.28 & 1.28 & 0.0 + 128 & 8.83 & 0.15 & 0.28 & 0.32 & 0.0 + 133 & 2.03 & 0.29 & @xmath1130.25 & 0.26 & 1.6 + 136 & 8.65 & 0.28 & 0.06 & & + 140 & 4.58 & 0.17 & @xmath1130.19 & @xmath1130.29 & 3.8 + rcrrcc 174 & g2 & 29.8 & 29.4 & const & 0 + 175 & g1 & 29.4 & 29.8 & ltvar & 0 + 175 & g2 & 30.3 & 29.8 & ltvar & 1 + 177 & g2 & 30.0 & 30.2 & flare & 0 + 177 & g1 & 30.3 & 30.2 & flare & 1 + 178 & t & 30.1 & & posfl & 1 + 187 & g2 & 30.5 & 30.3 & flare & 1 + 187 & g1 & 30.4 & 30.3 & flare & 0 + 188 & g2 & 29.5 & 29.7 & posfl & 0 + 188 & g1 & 30.6 & 29.7 & posfl & 1 + | we re - analyze all archival _ chandra / acis _ observations of the orion nebula cluster ( onc ) to study the x - ray properties of a large sample of pre main - sequence ( pms ) stars with optically determined rotation periods .
our goal is to elucidate the origins of x - rays in pms stars by seeking out connections between the x - rays and the mechanisms most likely driving their production rotation and accretion .
stars in our sample have @xmath0 near , but below , the saturation " value of @xmath1 . in addition , in this sample x - ray luminosity is significantly correlated with stellar rotation , in the sense of _ decreasing _ @xmath0 with more rapid rotation .
these findings suggest that stars with optical rotation periods are in the super - saturated " regime of the rotation - activity relationship , consistent with their rossby numbers .
however , we also find that stars with optical rotation periods are significantly biased to high @xmath2 .
this is not the result of magnitude bias in the optical rotation - period sample but rather to the diminishingly small amplitude of optical variations in stars with low @xmath2 .
evidently , there exists in the onc a population of stars whose rotation periods are unknown and that possess lower average x - ray luminosities than those of stars with known rotation periods .
these stars may sample the linear regime of the rotation - activity relationship .
accretion also manifests itself in x - rays , though in a somewhat counterintuitive fashion : while stars with spectroscopic signatures of accretion show _ harder _ x - ray spectra than non - accretors , they show _ lower _ x - ray luminosities and no enhancement of x - ray variability .
we interpret these findings in terms of a common origin for the x - ray emission observed from both accreting and non - accreting stars , with the x - rays from accreting stars simply being attenuated by magnetospheric accretion columns .
this suggests that x - rays from pms stars have their origins primarily in chromospheres , not accretion . |
thermal convection in a rotating spherical fluid shell represents one of the fundamental systems on which a considerable part of the understanding of many observed geophysical , planetary and astrophysical phenomena is based .
it is generally believed that the magnetic field exhibited by celestial bodies is generated in their interiors by convection driven by thermal and compositional buoyancy .
cloud patterns and the differential rotation seen at the surface of the major planets are likely to be connected with convection in the deep atmospheres of those planets ( busse , 1976 , 1994 ) .
considerable attention has therefore been devoted to the study of dynamical properties of rotating spherical shells . for recent reviews we refer to the papers by zhang and busse ( 1998 ) and busse ( 2002 ) . because a radially directed buoyancy force can not be easily realized in earth - bound laboratories , experimenters had to resort to the use of centrifugal accelerations ( busse and carrigan , 1976 ) or use spacecrafts for their experiments ( hart _ et al .
fortunately , it has turned out that the component of gravity perpendicular to the axis of rotation is much more important than the component parallel to it .
the centrifugal force is thus a convenient substitute since through the reversal of the applied temperature gradient the same buoyancy can be obtained as for the inward directed component of gravity perpendicular to the axis .
moreover , the centrifugal force increases with the distance from the axis just as the case of the gravity field of a spherical body of constant density . the use of electric fields ( hart _ et al .
_ 1986 ) or magnetic fields with ferrofluids ( rosenzweig _ et al .
_ 1999 ) leads to rather different radial dependences of the effective gravity .
it must be admitted , however , that the absence of an axial component of gravity in experiments employing the centrifugal force leads to a thermal wind type flow in the basic state . but this can be minimized when the coriolis parameter is sufficiently large or when a thin spherical shell is used as for example in the experimental realization of convection shown in figure [ f1 ] . in this
paper results from a numerical study will be reported that have been obtained with essentially the same computer code as employed in earlier studies focusing on turbulent convection ( tilgner and busse , 1997 ) and on the generation of magnetic fields ( grote and busse , 2001 ) . the mathematical formulation and the description of the numerical method given in section 2
can thus be kept brief . in section 3 a short introduction to the results of linear theory for the onset of convection
will be given . in section 4
properties of columnar convection at prandtl numbers of the order unity will be described and some animations of the time dependence will be presented .
equatorially attached convection flows are the subject of section 5 .
their properties at finite amplitudes are studied here for the first time .
because of the required high numerical resolution in time as well as in space the low prandtl number regime of fully nonlinear convection has posed a significant computational challenge until now .
as will be shown the continuity of the convection heat flux requires that equatorially attached convection must be joined by columnar convection as the rayleigh number increases significantly above its critical value . a concluding discussion will be given in section 6 .
for the description of finite amplitude convection in rotating spherical fluid shells we follow the standard formulation used by many authors zhang and busse(1987 ) , ardes _
( 1997 ) , zhang(1991 , 1992 ) .
it is assumed that a general static state exists with the temperature distribution @xmath0 where @xmath1 denotes the ratio of inner to outer radius of the shell and @xmath2 is its thickness .
@xmath3 is the temperature difference between the boundaries in the special case @xmath4 of vanishing internal heat sources .
the gravity field is given by @xmath5 where @xmath6 is the position vector with respect to the center of the sphere and @xmath7 is its length measured in units of @xmath2 .
in addition to the length @xmath2 , the time @xmath8 and the temperature @xmath9 are used as scales for the dimensionless description of the problem where @xmath10 denotes the kinematic viscosity of the fluid and @xmath11 its thermal diffusivity .
the density is assumed to be constant except in the gravity term where its temperature dependence given by @xmath12 const .
is taken into account .
the basic equations of motion and the heat equation for the deviation @xmath13 from the static temperature distribution are thus given by @xmath14 @xmath15 @xmath16 where the rayleigh numbers @xmath17 and @xmath18 , the coriolis parameter @xmath19 and the prandtl number @xmath20 are defined by @xmath21 since the velocity field @xmath22 is solenoidal the general representation in terms of poloidal and toroidal components can be used , @xmath23 by multiplying the ( curl)@xmath24 and the curl of equation ( [ 1a ] ) by @xmath6 we obtain two equations for @xmath25 and @xmath26 , @xmath27 \nabla^2 v + \tau q w - l_2 \theta = - { \mbox{\boldmath $ r$ } } \cdot \nabla \times [ \nabla \times ( { \mbox{\boldmath $ u$ } } \cdot \nabla { \mbox{\boldmath $ u$ } } ) ] \ ] ] @xmath28 w - \tau qv = { \mbox{\boldmath $ r$ } } \cdot \nabla \times ( { \mbox{\boldmath $ u$ } } \cdot \nabla { \mbox{\boldmath $ u$}})\ ] ] where @xmath29 and @xmath30 denote the partial derivatives with respect to time @xmath31 and with respect to the angle @xmath32 of a spherical system of coordinates @xmath33 and where the operators @xmath34 and @xmath35 are defined by @xmath36 @xmath37 stress - free boundaries with fixed temperatures are most often assumed , @xmath38 @xmath39 but sometimes no - slip boundaries are used in which case the conditions @xmath40 must be replaced by @xmath41 . for simplicity we shall assume the previously used value @xmath42 unless indicated otherwise .
the numerical integration of equations ( [ 1c ] ) and ( 2 ) together with boundary conditions ( [ 4 ] ) proceeds with the pseudo - spectral method which is based on an expansion of all dependent variables in spherical harmonics for the @xmath43-dependences , i.e. @xmath44 and analogous expressions for the other variables , @xmath26 and @xmath13 .
@xmath45 denotes the associated legendre functions . for
the @xmath7-dependence expansions in chebychev polynomials are used .
spherical harmonics up to the order of @xmath46 and up to 41 collocation points in radial direction have been used .
a time step as low as @xmath47 was required for the computations of the small prandtl number cases reported in section 5 . for further details
see busse _ et al . _
( 1998 ) , tilgner and busse ( 1997 ) .
benchmark comparisons for stationary solutions of equations ( [ 1c ] ) and ( 2 ) together with conditions ( [ 4 ] ) when the dynamo process is included have been done in the past ( christensen _ et al . _ ,
a rough idea of the dependence of the critical rayleigh number @xmath48 on the parameters of the problem in the case of internal heating can be gained from the expressions derived from the annulus model ( busse , 1970 ) @xmath49 @xmath50 @xmath51 where @xmath52 refers to the mean radius of the fluid shell , @xmath53 , and @xmath54 to the corresponding colatitude , @xmath55 arcsin@xmath56 .
the azimuthal wavenumber of the preferred mode is denoted by @xmath57 and the corresponding angular velocity of the drift of the convection columns in the prograde direction is given by @xmath58 . in figure [ f2 ]
the expressions ( [ 6a ] , [ 6c ] ) are compared with accurate numerical values which indicate that the general trend is well represented by expressions ( [ 6a ] , [ 6c ] ) .
the same property holds for @xmath57 . for a rigorous asymptotic analysis including the radial dependence we refer to jones _
et al . _ ( 2000 ) . in the following
we shall continue to restrict the attention to the case @xmath59 and thus we shall drop the subscript @xmath60 of @xmath17 .
there is a second mode of convection which becomes preferred at onset for sufficiently low prandtl numbers .
it is characterized by convection cells attached to the equatorial part of the outer boundary not unlike the `` banana cells '' seen in the narrow gap experiment of figure [ f1 ] .
the equatorially attached mode actually represents an inertial wave modified by the effects of viscous dissipation and thermal buoyancy .
an analytical description of this type of convection can thus be attained through the introduction of viscous friction and buoyancy as perturbations as has been done by zhang ( 1994 ) for the case of stress - free as well as for no - slip boundaries ( zhang , 1995 ) . according to ardes _
( 1997 ) equatorially attached convection is preferred at onset for @xmath61 where @xmath62 increases in proportion to @xmath63 . in the derivation of the asymptotic relationships ( 6 ) the exact nature of the boundary condition for the tangential component of the velocity vector at the spherical surface does not enter . in the limit of infinite @xmath19
the conditions of onset of convections are the same for no - slip and for stress - free conditions . in the following
we shall restrict the attention , however , to the case of stress - free boundaries in order to avoid the complications arising from ekman boundary layers at no - slip surfaces .
in general the onset of convection in rotating fluid spheres occurs supercritically .
as long as the convection assumes the form of shape preserving travelling thermal rossby waves as described by linear theory , its azimuthally averaged properties are time independent . in fact , as seen from a frame of reference drifting together with the convection columns the entire pattern is steady .
a differential rotation is generated through the action of the reynolds stress .
the latter is caused by the spiralling cross section of the columns which persists as a dominant feature at moderate prandtl numbers far into the turbulent regime .
the plots of the streamlines @xmath64 in the equatorial plane shown in figure [ f3 ] give a good impression of the spiralling nature of the columns .
a true time dependence of convection develops in the form of vacillations after a subsequent bifurcation .
first the transition to amplitude vacillations occurs in which case just the amplitude of convection varies periodically in time as exhibited in the left plot of figure [ f3 ] . at a somewhat higher rayleigh number shape vacillations
become noticeable which are characterized by periodic changes in the structure of the columns as shown in the right plot of figure [ f3 ] .
the outer part of the columns is stretched out , breaks off and decays .
the thinning and subsequent breakup can actually be best seen in movie 1 .
the tendency towards breakup is caused by the fact that the local frequency of propagation varies with distance from the axis according to expression ( 6c ) after @xmath54 has been replaced by the local colatitude @xmath65 .
+ movies 1 , 2 and 3 .
the two types of vacillations also differ significantly in their frequencies of oscillation .
this is evident from the time records of the energy densities of convection which have been plotted in figure [ f4 ] .
the various components of the energy densities are defined by @xmath66 @xmath67 where @xmath68 refers to the azimuthally averaged component of @xmath25 and @xmath69 is given by @xmath70 . as the rayleigh number
is increased further a fairly sudden transition into a chaotic regime occurs where convection has become strongly inhomogeneous in space and in time .
a typical sequence of plots is shown in figure 5 which covers about one period of the the relaxation cycles seen in the time record for @xmath71 in figure [ f4 ] .
in contrast to the more common relaxation oscillations encountered at higher rayleigh numbers , see , for example , the time record for @xmath72 in figure [ f4 ] , convection does not die off entirely at any time during the cycle .
but the interaction between convection and differential rotation appears to be similar .
as the amplitude of convection as measured by @xmath73 and @xmath74 grows the differential rotation generated by the reynolds stress grows as well with just a small delay in time .
when the differential rotation reaches a critical level the convection columns become disrupted and their amplitude decays .
subsequently the differential rotation decays as well on the time scale of viscous diffusion .
it is typical for this type of relaxation cycles that the viscous decay is shorter than the growth time of the differential rotation in contrast to the relaxation oscillations at higher values of @xmath75 .
the regime of relaxation oscillations is interrupted once in a while by a return to the more regular regime of shape vacillations as shown in the inserted enlargement of the time record for @xmath71 in figure [ f4 ] .
but in contrast to the vacillation of the right plot figure [ f3 ] the pattern is now strongly modulated .
the component with the azimuthal wavenumber m=1 plays a dominant role in this modulation .
as can be seen in figure 6 the pattern recurs nearly periodically in time in spite of the modulation .
figure 6 corresponds to the slightly higher rayleigh number of @xmath76 where convection changes intermittently between relaxation cycles and vacillation oscillations . as @xmath75
is further increased the spatio - temporal structure of convection becomes more irregular as can be seen in the time series for @xmath77 in figure [ f4 ] .
but at @xmath72 the relaxation oscillations with only intermittent convection become firmly established and they continue to persist up to rayleigh numbers of the order of @xmath78 .
they are basically the same phenomenon as found by grote and busse ( 2001 ) at @xmath79 and even the period is the same , about @xmath80 , which corresponds to the viscous decay time of the differential rotation .
the only difference that can be noted is the small precursor hump before the burst of convection . in figure 7
the third plot shows convection appearing in a fairly regular pattern .
afterwards it decays again as shown in the fourth plot before it erupts in a chaotic fashion as the differential rotation has reached its minimum amplitude . since this latter part of the cycle occurs on a rather short time scale , smaller intervals between the plots
have been used for this part in figure 7 .
the main difference between convection at @xmath81 and prandtl numbers of the order unity occurs at the transition between regular and irregular patterns .
a typical scenario is shown in figure 8 .
after convection has set in in the form of eight drifting column pairs the usual amplitude vacillations occur as the rayleigh number is increased .
the shape vacillations , however , exhibit a modulation with wavenumber @xmath82 as shown in figure 9 .
only every second pair of columns gets stretched until the outer part separates , a little earlier for the cyclonic column than for the anticyclonic one .
then the same process is repeated for the other pairs of columns such that the sequence shown in figure 9 exhibits only half a period of the oscillation .
in addition , of course , the column pattern drifts in the prograde direction .
movie 2 provides a vivid impression of this type of convection . with increasing rayleigh number
the stretching process gets out of phase and an @xmath83-modulation enters as shown in figure 10 .
the time dependence is still periodic , but has become more complex in that some of the separated outer parts become attached to the preceding column pair . to show this process in more detail some extra plots
have been added to the time sequence of figure 10 .
while the amplitude of convection varies considerably throughout the cycle , the differential rotation exhibits only small oscillations as can be seen from the corresponding section of figure 8 . by the time when @xmath75 has reached @xmath84
the modulated vacillations have become aperiodic and with increasing rayleigh number convection becomes more and more chaotic .
regularity reappears only in the form relaxation oscillations as shown in the section for @xmath85 of figure 8 .
remainders of the vacillations can still be seen in this time record .
but at @xmath86 the relaxation oscillations have become fully established with convection occurring only in a intermittent fashion .
although obtained for a slightly smaller value of @xmath19 , movie 3 gives a good impression of the onset of relaxation oscillations .
since stellar interiors as well as metallic planetary cores are characterized by rather small prandtl numbers much attention has been focused on the problem of the onset of convection in rotating fluid spheres in the case of small @xmath20 .
zhang and busse ( 1987 ) found the equatorially attached mode which is quite distinct from the columnar mode discussed in the preceding section .
the new mode represents an inertial oscillation which becomes excited when viscous dissipation is balanced by the energy provided by thermal buoyancy .
the fact that both energies can be regarded as small perturbations has led zhang ( 1994 , 1995 ) to solve the problem of onset of convection by an asymptotic analysis .
an detailed numerical study together with some analytical approximations can be found in the paper of ardes _
the results of these various efforts have turned out to be rather complex since the preferred inertial modes travel in the prograde as well in the retrograde directions depending on the parameters of the problem . moreover , the azimuthal wavenumber @xmath87 of convection does not increase monotonically with the coriolis parameter @xmath19 as is usually found for the columnar mode at values of @xmath20 of the order unity or higher .
in addition to the simple `` single cell '' inertial modes of zhang and busse ( 1987 ) multicellular modes have been found in the study of ardes _
( 1997 ) which appear to be closely related to the multicellular modes described by the airy function in the analysis by yano ( 1992 ) of the analogous problem of convection in the rotating cylindrical annulus .
the numerical solutions for the problem obtained by pino _
et al . _ ( 2000 ) also indicate the onset of multicellular convection in parts of the parameter space . here
we shall present only some examples of the variety of patterns encountered in a rotating sphere of a low prandtl number fluid . in figure 11 the multiple changes from retrograde to prograde modes as a function of @xmath19
are indicated and the corresponding changes in the patterns are shown in figure 12 . to facilitate comparisons the wavenumber
@xmath88 has been kept fixed . but
similar changes in the structure of convection are found when competing values of @xmath87 are admitted .
the retrograde and prograde modes differ little in their form .
the opposite phase between toroidal and poloidal components of motion is the most characteristic difference as indicated in figure 13 .
the sense of outward spiralling is also opposite for prograde and retrograde modes as is evident from figure 12 .
but in the limit of vanishing prandtl number the sense of spiralling disappears because the phase of the inertial modes does not vary with distance from the axis . whenever @xmath19 is sufficiently large the frequency omega is closely approximated by the frequency of the corresponding inertial modes which is given by the analytical expression ( zhang , 1994 ; ardes _ et al .
_ , 1997 ) , @xmath89^{\frac{1}{2}}).\ ] ] the negative ( positive ) sign applies for modes drifting in the prograde ( retrograde ) direction .
explicit values of expression ( 9 ) are given in figures 11 and 13 for comparisons with the numerically determined values .
for an analysis of nonlinear properties of equatorially attached convection we focus on the case @xmath90 with @xmath91 .
the critical rayleigh number for this case is @xmath92 corresponding to @xmath93 .
as r is increased beyond the critical value other values of @xmath87 from 7 to 12 can be realized , but @xmath93 and lower values are usually preferred .
an asymptotic perfectly periodic solution with @xmath93 or @xmath94 can be found only for rayleigh numbers close to the critical value when computations are started from arbitrary initial conditions . on the other hand ,
perfect periodic patterns appear to be stable with respect to small disturbances over a more extended regime of supercritical rayleigh numbers .
distinct transitions like the transition to amplitude vacillations and to structure vacillations do not seem to exist for equatorially attached convection .
instead modulated patterns are typically already observed when @xmath75 exceeds the critical value by 10% as can be seen in the plots of figure 15 .
these modulations are basically caused by the interaction of two modes with neighboring values of the azimuthal wavenumber @xmath87 as indicated in figure 16 .
here the region of small amplitude drifts in the retrograde direction with the difference of the drift rates for @xmath95 and @xmath88 .
the time series of energy densities shown in figure 17 indicate that more than two modes usually contribute to the dynamics of the pattern since the time dependence is not periodic as one would expect if only two modes interact .
the computations of the time series require a high spatial resolution together with a small time step .
the time spans indicated in figure 17 are sufficient for reaching a statistically steady state of the fluctuating components of motion since these equilibrate on the fast thermal time scale of the order @xmath96 . only close to @xmath97
the adjustment process takes longer as can be seen in the case @xmath98 where a @xmath93 - pattern approaches its equilibrium state .
the pattern corresponding to the other cases of figure 17 are the ones shown in figures 15 , 16 and 18 .
the differential rotation represented by @xmath99 relaxes on the viscous time scale and thus takes a long time to reach its asymptotic regime in the examples shown in figure 17 .
but the differential rotation is quite weak such that it has a negligible effect on the other components of motion except in the case of the highest rayleigh number of figure 17 .
even smaller is the axisymmetric part of the poloidal component of motion which is not shown in the plots of figure 17 . at higher rayleigh numbers the equatorially attached eddies spread farther into the interior and in some cases
become detached from the equator as can be seen in the plots of figure 18 . in this way the convection eddies contribute to the heat transport from the inner boundary .
but at the same time they acquire the properties of the convection columns which are characteristic for convection at higher prandtl numbers . accordingly
the differential rotation is steeply increased at @xmath100 and a tendency towards relaxation oscillation can be noticed in the upper right time series of figure 17 .
the results on linear and nonlinear properties of low prandtl number convection in rotating spherical shells are of special interest for applications to problems of convection in planetary and stellar interiors .
since planetary cores often consist of liquid metal or of metallic hydrogen as in the case of jupiter and saturn , their prandtl numbers are rather low . moreover , because of high temperatures radiative heat transport is no longer negligible which tends to lower the prandtl number even more .
the latter effect dominates in stellar interiors .
on the other hand , turbulence owing to small scale motions which remain unresolved in numerical simulations will tend to equalize all effective diffusivities .
but this tendency is not likely to erase entirely the differences in the subgrid scale transports of heat and momentum . of particular interest
is the effect of the prandtl number on the dynamo action of convection .
while the strong differential rotation associated with columnar convection tends to promote the generation of magnetic fields ( see , for example , grote and busse , 2001 ) , its near absence in the case of equatorial attached convection may increase the rayleigh number required for dynamo action at a given value of the magnetic prandtl number . as the rayleigh number reaches high values more column like
convection flows tend to appear near the inner boundary of the shell and are expected to promote dynamo action .
numerical studies of this problem are in progress and will be reported in the near future .
the regular and modulated patterns of convection reported in this paper as well as the coherent structures found in the chaotic regimes of convection can in principle be observed in experiments through the use of techniques mentioned in the introduction .
low prandtl number fluids such as liquid metals pose severe experimental challenges because convection flows can not easily be visualized . on the other hand
the temperature profiles of convection patterns can be measured through time records of thermistor probes implanted in the spherical boundary since the pattern is drifting by .
see , for example , measurements of convection in mercury in the case of the cylindrical annulus by azouni _
( 1986 ) and of convection in a spherical shell by cordero and busse ( 1992 ) .
for the purpose of an eventual comparison with experimental observation the computations reported in this paper should be repeated for the case of no - slip boundaries with @xmath101 and @xmath102 .
the computational effort will be more demanding but no significant qualitative changes are expected in that case .
* ardes , m. , busse , f.h . , and wicht , j. , thermal convection in rotating spherical shells , _ phys . earth plan .
int . _ * 99 * , 55 - 67 , 1997 * azouni , a. , bolton , e.w
. , and busse , f.h . , experimental study of convection in a rotating cylindrical annulus , _ geophys .
fluid dyn .
_ , * 34 * , 301 - 317 , 1986 * busse , f.h . : thermal instabilities in rapidly rotating systems .
_ j. fluid mech .
_ , * 44 * , 441 - 460 , 1970 * busse , f.h . , a simple model of convection in the jovian atmosphere , _ icarus _ * 20 * , 255 - 260 , 1976 * busse , f.h . , convection driven zonal flows and vortices in the major planets , _ chaos _ , * 4 * , 123 - 134 , 1994 * busse , f.h . , convection flows in rapidly rotating spheres and their dynamo action , _ phys . fluids _ , * 14 * , 1301 - 1314 , 2002 * busse , f.h . , carrigan , c.r . , laboratory simulation of thermal convection in rotating planets and stars , _ science _ , * 191 * , 81 - 83 , 1976 * busse , f.h . ,
grote , e. , tilgner , a. , on convection driven dynamos in rotating spherical shells .
_ studia geoph .
_ , * 42 * , 211 - 223 , 1998 * christensen , u.r . _
et al . _ , a numerical dynamo benchmark . _
phys . earth plan .
_ , * 128 * , 25 - 34 , 2001 * cordero , s. , and busse , f.h .
, experiments on convection in rotating hemispherical shells : transition to a quasi - periodic state , _ geophys . res .
_ , * 19 * , 733 - 736 , 1992 * grote , e. , and busse , f.h . , dynamics of convection and dynamos in rotating spherical fluid shells , _
fluid dyn .
_ , * 28 * , 349 - 368 , 2001 * hart , j.e . , glatzmaier , g.a.,toomre , j. , space - laboratory and numerical simulations of thermal convection in a rotating hemi - spherical shell with radial gravity , _ j. fluid mech .
_ , * 173 * , 519 - 544 , 1986 * jones , c.a . , soward , a.m. , and mussa , a.i . ,
the onset of thermal convection in a rapidly rotating sphere , _ j. fluid mech .
_ , * 405 * , 157 - 179 , 2000 * pino , d. , mercader , i. , and net , m. , thermal and inertial mode of convection in a rapidly rotating annulus , _ phys
e _ , * 61 * , 1507 - 1517 , 2000 * rosenzweig , r.e . ,
browaeys , j. , bacri , j .- c . , zebib , a. , and perzynski , r. , laboratory study of spherical convection in simulated central gravity , _ phys .
letts .. _ , * 83 * , 4904 - 4907 , 1997 * tilgner , a. , and busse , f.h .
, finite amplitude convection in rotating spherical fluid shells , _ j. fluid mech .
_ , * 332 * , 359 - 376 , 1997 * yano , j .-
, asymptotic theory of thermal convection in rapidly rotating systems , _
j.fluid mech._,*243 * , 103 - 131 , 1992 * zhang , k. , vacillating convection in a rotating spherical fluid shell at infinite prandtl number , _ j. fluid mech . _ , * 228 * , 607 - 628 , 1991 * zhang , k. , spiraling columnar convection in rapidly rotating spherical fluid shells ,
_ j. fluid mech .
_ , * 236 * , 535 - 556 , 1992 * zhang , k. , on coupling between the poincar equation and the heat equation , _ j. fluid mech .
_ , * 268 * , 211 - 229 , 1994 * zhang , k. , on coupling between the poincar equation and the heat equation : no - slip boundary condition , _ j. fluid mech . _ , * 284 * , 239 - 256 , 1995 * zhang , k. , and busse , f.h . , on the onset of convection in rotating spherical shells , _ geophys . astrophys .
fluid dyn .
_ , * 39 * , 119 - 147 , 1987 * zhang , k. , and busse , f.h . , some recent developments in the theory of convection in rotating systems , in `` nonlinear instability , chaos and turbulence '' , vol 1 , l. debnath and d.n .
riahi , eds . , _
computational mechanics publications , wit press _ , 17 - 69 , 1998 | new j. phys . , 5 , pp . 97.1 - 97.20 , doi:10.1088/1367 - 2630/5/1/397 , 2003
. + patterns of convection in internally heated , self gravitating rotating spherical fluid shells are investigated through numerical simulations .
while turbulent states are of primary interest in planetary and stellar applications the present paper emphasizes more regular dynamical features at rayleigh numbers not far above threshold which are similar to those which might be observed in laboratory or space experiments .
amplitude vacillations and spatial modulations of convection columns are common features at moderate and large prandtl numbers . in the low prandtl number regime
equatorially attached convection evolves differently with increasing rayleigh number and exhibits an early transition into a chaotic state .
relationships of the dynamical features to coherent structures in fully turbulent convection states are emphasized . |
the outer disks of galaxies are increasingly being recognized as a diverse environment whose properties may aid us in understanding the formation of galaxies . since the radio observations of the 1950s @xcite
, we have known that the baryonic component of spiral galaxies , as traced by neutral hydrogen , is significantly larger than suggested by classical size indicators such as @xmath1 ( the radius where the surface brightness drops to 25 mag arcsec@xmath3 ; we use measurements of the @xmath4-band surface brightness obtained from the literature throughout this analysis ) .
optical studies , however , have focused on the inner parts of galaxies , where , among other aspects , the radial distribution of h@xmath0 emission has been subject of intense scrutiny for many years .
for example , @xcite used narrow - band imaging of nearby galaxies to construct h@xmath0 surface brightness profiles .
such profiles usually exhibit a truncation " in the h@xmath0 surface brightness at relatively large radii , although usually still within the optical radius @xmath1 .
this truncation is also reflected in the scarcity of h@xmath0 rotation curves that extend to radii much beyond @xmath5 in comprehensive long - slit surveys ( e.g. , @xcite , being one of the deepest such surveys , observe no rotation curves beyond @xmath1 in a sample of 329 ) . a widely accepted hypothesis for such a truncation is the existence of a threshold in the surface gas density , below which star formation becomes inefficient @xcite .
alternate explanations include the hypothesis that the truncation is indicative of an actual break in the mass distribution that is related to the initial formation conditions of the disk @xcite . however ,
truncation " may be too strong a descriptive statement because it is now known that low - level star formation occurs at large galactocentric radii , well beyond the radius corresponding to the critical threshold . @xcite have detected individual h@xmath0 emission regions in the outer disks of spiral galaxies as far as 2 @xmath1 , and uv emission appears relatively common in outer disks @xcite .
furthermore , in a study of ngc 3814 using deep two - band optical imaging , @xcite have found statistically significant overdensities of marginally resolved sources in the outer disk that are likely to be star clusters .
this outer - disk star formation has raised new interest in the hypothesis of a star formation threshold .
the absence of evidence for a break in the surface brightness profiles of uv emission , which is also a star formation indicator , led @xcite to suggest that the star formation rate does not exhibit a break , and that the h@xmath0 truncation is a stochastic effect . as the expectation number of star formation regions with stars massive enough to generate a stroemgren sphere drops below unity
, the h@xmath0 emission profile goes to zero , while low level(mass ) sf continues as measured with the uv .
other , perhaps more exciting , possible explanations include a change in the initial mass function @xcite .
these arguments depend critically on reliable measures of h@xmath0 and continuum profiles out to large radii . at the same time
, studies of faint optical continuum emission have shown that stellar disks can also be traced far into the outer disk .
while some galaxies exhibit a single exponential surface brightness profile to the largest measurable radii @xcite , others are more accurately described by broken exponential profiles with a characteristic break radius , beyond which the stellar surface brightness profile may be shallower or steeper than in the inner disk @xcite .
these profiles are sometimes described as exponential , sub - exponential , and super - exponential @xcite ; the exponential and sub - exponential types are also variously referred to as freeman types i and ii @xcite , and the designation type iii " has come into usage to describe the super - exponential ( up - bending ) shape .
how these outer stellar structures are related to possible structures in the gaseous disk , and in particular , h@xmath0 emission , is a crucial question in understanding the origin of the outer disk stellar populations . is
the truncation in h@xmath0 emission a real indicator of a truncation in star formation , or , as @xcite suggest , merely a stochastic effect ?
if , on the other hand , the break in the star formation surface density is real , is it also responsible for the characteristic break in the stellar continuum profile ? in that case , is there significant star formation beyond the break , and is it enough to have created the outer disk stellar content in situ ? or must other processes be invoked to populate the outer disks with stars ? to answer these questions , a systematic and quantitative study of the distribution of h@xmath0 emission at large radii is required .
the traditional technique of narrow - band imaging with subsequent subtraction of broad - band continuum emission is generally insufficiently sensitive to probe to large radii .
the limitation lies not only in the achievable signal - to - noise ratio ( which is limited because typical narrow - band filters are much wider than the h@xmath0 emission line ) , but also in the stellar continuum subtraction . at large radii , because the spectral flux density at the peak of the h@xmath0 line is of the same order of magnitude as the stellar continuum , the wavelength dependence of the stellar continuum spectral energy is sufficient to significantly degrade the subtraction .
for example , absorption troughs , which the h@xmath0 emission line is often embedded in , can render it undetectable .
the way around these problems is to use higher spectral resolution .
this can involve very narrow filter bandpasses , which is now possible over a wide range of redshifts due to the increasing availability of tunable filters @xcite , or traditional spectroscopy . in a pioneering effort in the study of gaseous outer disks , @xcite detected h@xmath0 emission at @xmath6 r@xmath7 , beyond the truncation radius of the neutral hydrogen disk , using the fabry - perot staring technique .
however , true spectroscopic observations provide improvements in the subtraction of the stellar continuum , of the [ nii ] emission lines , and of h@xmath0 absorption troughs , allowing one to reach sensitivity limits fainter than 10@xmath8 erg s@xmath9 cm@xmath3 arcsec@xmath3 ( see also @xcite ) .
we have observed late - type , edge - on disk galaxies with multi - hour , long - slit spectroscopy . the long - slit technique , although suffering from the much lower throughput in comparison to the fabry - perot technique , allows us , for edge - on geometry , to take advantage of fields of view that cover the entire galaxy and greatly facilitates sky and continuum subtraction .
we typically detect h@xmath0 emission to galactocentric radii of 1.5 r@xmath7 , and in certain cases up to 2 r@xmath7 . in a previous paper @xcite
, we discuss the kinematic properties of these outer h@xmath0 disks and find the kinematics to be disk - like , with generally no indication of kinematic anomalies or higher velocity dispersions as one approaches the outer edge of the disk .
a subsample of galaxies with known optical warps has been studied separately @xcite to determine if kinematic anomalies , such as breaks in the rotation curve , were associated with the onset of warps that could indicate ongoing accretion processes of compact hi clouds or satellite galaxies as causes of the warp .
we use data from both of these samples in this paper .
we now turn our attention to the surface brightness profile of h@xmath0 emission in the outer disk .
our aim is to measure the distribution of h@xmath0 emission in the outer disk and determine if the h@xmath0 emission profiles can be categorized as done for the stellar continua .
we will determine whether there is indeed a break in the h@xmath0 surface brightness profile , and if so , whether this break can be associated with that in the stellar surface brightness . if there is a break , is it steep enough to constitute a truncation ? and
how is the stellar continuum , a measure of integrated star formation , distributed in comparison to the current star formation , as indicated by h@xmath0 ?
we describe our sample in
[ sec_data ] . in
[ sec_composite ] , we discuss and compare the properties of h@xmath0 and stellar continuum emission based on a composite of our 15 individual galaxy spectra , while in [ sec_individual ] , we discuss what constraints we can place on the profile shapes of individual galaxies .
our sample comes from three observing runs : a three - night run using fors1 at the vlt in november 2004 , a three - night run with gmos at gemini - south in april 2005 , and a three - night run using fors2 at the vlt in september 2007 .
target selection was , in all cases , for isolated , edge - on , late - type galaxies of several arcminutes in diameter , preferentially at a redshift of several thousand km / s , which places the h@xmath0 line in a relatively quiet window of the sky background .
an important difference among the runs is that the targets for the first two runs were chosen to be as morphologically undisturbed and regular as possible , whereas targets for the third run deliberately included several objects with optical warps , mostly taken from the catalog by @xcite . at first
, it may seem incongruous to include a set of warped galaxies in a long - slit experiment , because the disk material is eventually going to curve away from the slit at the onset of the warp .
however , given that nearly all galaxies might be warped to some extend @xcite , even if they appear normal " in the available imaging , it is useful to include a subsample of known warped galaxies to provide the internal control against which to view our results . furthermore , warps typically set in around the @xmath1 radius @xcite , so that several important results from the undisturbed sample within this radial distance may still be verified with the warped sample .
all 15 targets are listed in table [ tab_sample ] .
the table lists the r@xmath7 radii adopted for this analysis , summarizes which telescope and instrument were used , and identifies whether the galaxy is known to be warped or not .
values of r@xmath7 are drawn from @xcite for all objects from the fors1 and gmos - s runs and from the third reference catalogue of bright galaxies @xcite for all objects from the fors2 run .
the projected optical diameter , @xmath11 , of all galaxies is at most half the slit length , so that each object is covered completely by the slit in a single observation , along with a substantial sky sample . in our analysis , we group the galaxies into an undisturbed and a warped sample .
two of the `` unwarped '' galaxies ( eso 340-g026 , mcg -01 - 10 - 035 ) were observed during the september 2007 run , which was primarily targeting warped objects .
we describe the surface brightness profile as a function of projected radial distance from the galaxy center in units of the r@xmath7 radius so as to permit a comparison between galaxies of different physical sizes .
other characteristic size measures , such as the half - light radius , @xmath12 , and the radius containing 90% of the total light , @xmath13 , were also considered .
however , all of these are generally strongly correlated ( for all galaxies in our sample and in @xcite , r@xmath14r@xmath15 with a scatter of 15@xmath16 , and with the exception of a single galaxy , r@xmath17r@xmath18 to within 15@xmath16 ) .
we choose to use r@xmath7 because it is the only such quantity available in the literature for all of our galaxies and because it is a robust measure for the `` edge '' of the disk in this sample where inner - disk size measures can be heavily affected by dust . [ cols="<,<,<,<,<,<",options="header " , ] + we now discuss the results in terms of specific questions .
hypothesis @xmath19 is that of an exponential profile with a sharp truncation .
most galaxies in our sample , whether warped or not , are inconsistent with the truncated profile shape .
the two cases that are marginally consistent with a sharp truncation are eso 478-g011 and ic 2058 .
visual inspection supports this for ic 2058 ; in the case of eso 478-g011 , the best - fit truncation radius is at the very large radius of @xmath20 , so that the truncated profile is essentially indistinguishable from a single exponential shape .
we now restrict this hypothesis . in many of our sample galaxies , h@xmath0 is observed to large radii , as table [ tab_outermost ] shows .
such detections are formally inconsistent with an exponential profile truncated at smaller radii . therefore , we impose the additional constraint that the truncation radius be located just beyond the radius of the outermost h@xmath0 detection and choose @xmath21 for this purpose . in table
[ tab_indresults ] , the results of this analysis are shown in column @xmath22 .
none of the galaxies in our sample are described well by a profile truncated at the last significant h@xmath0 detection .
lastly , perhaps a sharply truncated profile is ruled out by this analysis only because the truncation is softened in projection due to the nearly - edge - on configuration .
simple simulations of the projected profiles that would be obtained from such a configuration show that , even for an almost perfectly edge - on geometry , the softening effect towards the edge of the h@xmath0 disk would be fairly subtle and unlikely to exceed the surface brightness fluctuations observed in the real spectra .
therefore , this analysis confirms that a broken - exponential profile with a finite outer - disk slope provides a significantly better fit to the individual galaxy profiles than a sharply truncated single exponential law .
next , we consider @xmath23 , the hypothesis that the h@xmath0 surface brightness profile is described by a single , unbroken exponential .
this model corresponds to the freeman type i profile , which is observed to hold for the stellar continuum of some galaxies @xcite out to extremely large radii .
our sample is dominated by galaxies with broken exponential profiles in the stellar continuum ( figures [ fig_profiles1 ] to [ fig_profiles3 ] ) , where visual inspection identifies only few objects as being consistent with a single exponential profile .
for about half of the h@xmath0 profiles in our sample , whether they represent normal or warped galaxies , we also rule out the unbroken exponential profile shape ( @xmath23 ) , as table [ tab_indresults ] shows .
the exceptions among the unwarped galaxies may be of particular interest because the existence of a critical threshold surface gas density beyond which the specific star formation rate in a galaxy changes abruptly or even drops to zero has been suggested repeatedly in the literature @xcite .
the objects for which we can not conclusively rule out an unbroken exponential law are eso 299-g018 , eso 385-g008 , eso 478-g011 , ic2058 , mcg -01 - 10 - 035 , ngc 259 , and ugc 12423 . in some cases ,
this can be attributed to the strong fluctuations , which lower the quality of even the broken - exponential fit and reduce our ability to discriminate between different hypotheses .
visual inspection does single out two of these objects , eso 478-g011 and ngc 259 , as the best candidates for an unbroken exponential profile ; these are also the two objects best fitted under @xmath23 .
this is surprising for ngc 259 , which is listed as a warped galaxy ; however , within our warped sample , this object is the least - inclined , and the classification as a warped object may be due to a prominent spiral arm , rather than to an actual vertical bending of the disk , which may explain why we can trace a regular , possibly unbroken h@xmath0 profile out to @xmath24 .
we conclude that , just like the composite profile , the individual galaxy h@xmath0 profiles are generally inconsistent with unbroken exponentials out to large radii , but that a significant number of exceptions might exist in our sample . the third question we address is whether the individual galaxies might all share a common break radius , @xmath25 , as measured in the composite profile .
is this radius a universal property of late - type galaxies , or simply an average for a population with an intrinsically large scatter ?
we address this question by starting with h@xmath26 , and then fixing the break radius at @xmath27 , but leaving @xmath28 , @xmath29 , and @xmath30 as free parameters .
we refer to this model as h@xmath31 .
the result , presented in the @xmath32 column of table [ tab_indresults ] , shows that all but one ( eso 323-g033 ) of our galaxy profiles are consistent ( @xmath33 discrepant ) with broken exponential profiles with a break radius of @xmath27 .
eso 323-g033 bears the distinction of being consistent with neither of the four alternative profile shapes we are considering here . because a single inconsistency at the 2@xmath34 level in a sample of 15 is within statistical expectations ,
we conclude that the hypothesis of a universal " break radius for late - type galaxies like we have selected is consistent with these data . extending the previous analysis
, we now examine whether the composite profile calculated in [ sec_composite ] is itself an acceptable representation of all individual galaxies in the sample .
again , we create mock profiles , this time with the composite spectrum for which we allow a renormalization . the results ( table [ tab_indresults ] )
show that the h@xmath0 profiles of most of our galaxies are inconsistent with the composite ; there are only four exceptions : one from the undisturbed sample ( eso 299-g018 ) and three from the warped sample ( eso 340-g026 , mcg -01 - 10 - 035 , ugca 23 ) .
we conclude that there are real variations among the galaxies and that the composite spectrum is not necessarily a good representation of a galaxy chosen at random , with the exception of the break radius at @xmath35 .
we find that more than half of our galaxies do not have a freeman type i h@xmath0 surface brightness profile ( a single , unbroken exponential ) .
visual inspection demonstrates that the profiles are rather of the downbending ( sub - exponential ) type ii class , and are thus qualitatively following the behaviour of the stellar continuum , i.e. , their surface brightness profiles may be described as broken exponential profiles with a characteristic break radius and a steeper slope in the outer disk than in the inner disk .
the h@xmath0 break radius at @xmath36 r@xmath7 appears surprisingly well - constrained and is observable both in the unperturbed and the warped subsamples .
observations of a break in the h@xmath0 surface brightness profile are not new .
breaks or truncations in the h@xmath0 profile have been noted for many years ( cf . * ? ? ?
* ) , and a star formation threshold , i.e. , a critical gas surface sensitivity below which star formation can not proceed or proceed only very inefficiently has often been suggested as the most likely explanation . other breaks or truncations are expected at very large radii , where ionization by the cosmic ultraviolet background may become a significant contributor to outer - disk h@xmath0 emission and where the baryonic disk ends ; however , both of these effects are likely occur at or beyond 1.5 @xmath1 , which is a typical outer radius of hi disks , and at surface brightness levels fainter than what we can reliably probe with these data .
what is new here is that we quantify the radial behavior of the emission beyond this break radius and out to large radii , @xmath37 .
it is tempting to identify the clear break in h@xmath0 with the h@xmath0 , and possibly star formation , threshold well - known from the literature @xcite , but if so , then this threshold " is clearly not the end of star formation because h@xmath0 emission continues with some regularity beyond .
furthermore , if the threshold position depends on local physical conditions , it is remarkable how well - defined the location of @xmath38 is with respect to @xmath1 .
it is also tempting to identify the break radius in the stellar continuum that we observe with that observed by @xcite , and then further associate it with our observed h@xmath0 break radius because the two are consistent within 2@xmath34 in our data .
however , the outer - disk exponential slopes are different for the two components , with the stellar continuum being more extended , suggesting that there is not a 1:1 correspondence between the surface brightness profiles of the two components .
the physics determining the two are therefore more complex than a simple relationship between star formation and stellar populations .
if h@xmath0 is a linear tracer of star formation , the current stellar continuum surface brightness distribution can not have been produced by in - situ star formation , unless star formation was more extended in the past . because the currently favored paradigm is that galaxies form inside - out " @xcite , we consider it more likely that either the outer - disk stellar population has resulted from a redistribution of stellar mass ( for example , as a result of tidal interactions with other galaxies @xcite or secular dynamical evolution @xcite ) , or that h@xmath0 is an increasingly biased tracer of star formation towards large radii .
no less important than the observation of a break in the h@xmath0 surface brightness profile between the inner and the outer disk is the complementary observation that h@xmath0 emission is not sharply truncated
. we observe h@xmath0 emission as far out as @xmath39 , and find its surface brightness profile to continue with some regularity , possibly even following an exponential law , but with a steeper slope .
the observation of such extended emission is consistent with observations in recent years of extended uv emission around several galaxies @xcite .
it demonstrates that star formation can occur even in the extreme outer - disk regime , where gas surface densities are presumably very low . at the same time
, the presence of the break in the h@xmath0 profile suggests that , even if star formation is not completely suppressed in the outer disk , it appears to proceed under different physical conditions .
the finding that the outer disk uv emission , which is also an indicator of recent star formation , does not exhibit a break @xcite , has given rise to the suggestion that the star formation rate profile is also unbroken .
a possible explanation of the divergent behaviour between the h@xmath0 and uv profiles is that the truncation of h@xmath0 emission is a stochastic effect due to the number of star formation regions with sufficiently massive , ionizing stars dropping below unity @xcite .
however , in our observations , which reach much higher sensitivies than narrow - band imaging studies and superpose multiple galaxies , we detect emission at extremely high levels of confidence , at and beyond the hypothetical truncation radius . therefore , there are numerous ionizing sources in the outer disk .
nonetheless , we see a clearly - defined break in the h@xmath0 surface brightness profile both in individual galaxies and in the composite ; we conclude that the break in the h@xmath0 profile is real , but that it does not mark the end of star formation or the end of the formation of sufficiently massive stars . the apparent disagreement with the uv surface brightness profile is thus puzzling .
if we interpret the uv observations to mean that star formation is not subject to a sudden break , then h@xmath0 must be a biased tracer of star formation and suppressed at large radii .
the ratio of uv to h@xmath0 emission must change continuously with radius .
one potential cause of such an effect is an increasingly bottom - heavy initial mass function with radius , either as a result of a real change in the functional form or from statistical effects related to the increasingly poorer sampling of the upper stellar masses as total mass decreases .
a scenario in which the _ integrated galactic _ imf ( which is probed here ) may vary with environment as a result of variations of the mass function of molecular clouds across the galaxy , while the shape of the imf itself would be preserved , has been presented by @xcite .
another potential cause is a transition from ionization - bounded to density - bounded hii regions .
if a larger fraction of ionizing photons escape without being converted to h@xmath0 at larger radii the uv to h@xmath0 ratio would change . within inner galaxy disks
the effect seems to go the other way , with high - luminosity star - forming regions , rather than low - luminosity ones , appearing to be density bounded @xcite , but the situation in the outer disk could be quite different .
a final explanation for the increasing bias between stellar continuum and h@xmath0 emission in the outer disk is that outer - disk star formation may be an intermittent phenomenon , and short , but correlated , episodes of star formation in the past may have created the present extended stellar population before returning to the low - level state that we observe in our sample .
a similar scenario , involving bursts of star formation across the entire disk , has been proposed by @xcite based on galex observations of uv colors of outer galaxy disks . however
, this scenario does not appear to readily explain why the uv surface brightness profile would not exhibit a truncation , unless the galaxy was observed shortly after such a burst , so that the uv outer disk would still be dominated by stars formed during the recent star formation episode , while h@xmath0 emission , which traces star formation only over very short timescales , has already abated .
@xcite find that the stellar surface mass density also exhibits a much less pronounced break than the optical light of the stellar continuum .
they suggest that a change in the composition of the stellar population , rather than an actual break in the mass distribution , is responsible for the observed break in the optical continuum surface brightness . in this context
, it is surprising that both the stellar surface mass density and the integrated , recent star formation rate ( as measured by uv emission ) supposedly show no or only weak breaks , but both stellar continuum and h@xmath0 do . as exceptions to the rule ,
those galaxies where we find no evidence for a truncation or break in the h@xmath0 emission deserve particular attention , as any theory attempting to explain the presence of a break in the h@xmath0 profile for example , by invoking a critical gas surface density threshold below which star formation is inefficient must also be able to provide an explanation for objects that apparently violate the rule . of the objects that are formally consistent with an unbroken exponential profile , ngc 259 and
eso 478-g011 may be interesting cases for follow - up studies .
a final consideration must be given to the effect of dust on the h@xmath0 surface brightness profiles , and , in particular , to the question whether dust absorption could influence our measurement of the characteristic break radius and possibly even our classification of the composite h@xmath0 profile as a ( sub - exponential ) type ii , rather than a ( single exponential ) type i. if dust extinction sets in around this radius and increases continuously towards the galaxy center , absorbing an increasing fraction of h@xmath0 flux , it may artificially flatten the profile slope and possibly even create the false impression of a broken exponential shape .
this scenario , however would imply that the unabsorbed h@xmath0 profile could be recreated by extrapolating the outer - disk h@xmath0 profile back towards the center , implying that the real h@xmath0 surface brightness in the inner regions would be more than an order of magnitude higher than observed here .
as dust shielding is unlikely to be uniform ( even in cases of nearly perfect edge - on alignment , the slit intercepts regions both on and off the central dust lane ) , at least some sight lines should permit glimpses of these substantially higher surface brightness levels , leading to a larger dynamic range in the fluctuations of the inner - disk surface brightness profile .
this is not the case ; even the highest h@xmath0 peaks remain far below the levels expected from interpolating the outer - disk profile , confirming that the break is indeed real .
dust shielding may , however , plausibly modify the measured value of the inner - disk profile slope by smaller amounts . an additional argument against dust obscuring a large fraction of the h@xmath0 flux comes from the rotation curves .
if dust played a significant role up to the break radius of 0.7 @xmath1 , contributions to the observed h@xmath0 flux would be weighted towards gas on the outskirts of the galaxy along the edge closer to the observer .
we would therefore expect the rotation curve to rise very slowly in the inner disk and not flatten before it reaches this point .
however , most rotation curves in this sample are already flat or close to flat well before reaching 0.7 @xmath1 @xcite , indicating that the h@xmath0 emission at this point already samples the kinematics of the gas along the entire sight line , and that the observed kinematics and flux can not be strongly biased by internal dust extinction .
we have presented a study of h@xmath0 emission and stellar continuum in 15 low - redshift , late - type , edge - on galaxies .
the use of deep long - slit spectroscopy , rather than narrow - band imaging , allows us to probe h@xmath0 emission to very faint levels , @xmath4010@xmath8 erg s@xmath9 cm@xmath3 arcsec@xmath3 .
h@xmath0 emission is traced out to 50% beyond the r@xmath7 radius , roughly twice as far as typical h@xmath0 rotation curves in the literature .
it is thus possible to probe the outer galactic disks , which have hitherto primarily been a domain of radio astronomy , in h@xmath0 emission , and thus simultaneously gain information about kinematics , star formation , metallicity , and stellar continuum with arcsecond - scale spatial resolution . in the present paper , we have focused on the properties of the surface brightness profile of h@xmath0 emission and compared them to those of the stellar continuum .
our conclusions are : * in this sample , which is dominated by freeman type ii ( sub - exponential " ) profiles in the stellar continuum ( a broken exponential profile with a steeper slope in the outer than the inner disk ) , h@xmath0 emission in the composite profile also follows a type ii profile .
for the h@xmath0 composite profile , we can rule out both an unbroken single exponential profile as well as a truncated profile , for which the flux drops to zero at a given break radius .
* there is a well - defined break radius in the composite h@xmath0 surface brightness profile at @xmath25 r@xmath7 . *
the h@xmath0 and stellar continuum distributions may be consistent with a single break radius @xmath38 , but the stellar continuum profile drops more slowly in the outer disk . * the presence of a clearly defined break in the h@xmath0 surface brightness profile , despite the fact that our observations of the outer disk slope are clearly not limited by poisson noise , indicates that any apparent deficit of h@xmath0 emission in the outer disk is not , in general , a stochastic effect arising from low number statistics .
the break in h@xmath0 is real , but it is not a truncation , as evidenced by the common presence of low - level outer - disk h@xmath0 emission to @xmath41 . * for most objects in our sample , a truncated exponential as well as an unbroken exponential can be ruled out even on an individual basis , thereby requiring a broken exponential profile .
all but one of our galaxies are individually consistent with a broken exponential profile , with a fixed h@xmath0 break radius at @xmath35 , showing that the conclusions drawn from the composite profile describe not just a meaningless average , but are representative of at least a significant fraction of individual galaxies .
based on observations obtained at the gemini observatory ( program i d gs-2005a - c-4 ) , 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 ) .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
d.c . gratefully acknowledges financial support from the fundacin andes .
dz acknowledges that this research was supported in part by the national science foundation under grant no .
phy99 - 07949 during his visit to kitp , a guggenheim fellowship , generous support from the nyu physics department and center for cosmology and particle physics during his sabbatical there , nasa ltsa grant nng05ge82 g , nsf ast-0307482 , and the david and lucile packard foundation .
jbh is funded by a federation fellowship from the australian research council .
bakos , j. ; trujillo , i. ; pohlen , m. , 2008 apjl 683 , 103 beckman , j. e. ; rozas , m. ; zurita , a. ; watson , r. a. ; knapen , j. h. ; 2000 aj 119 ,
2728 bland - hawthorn , j. , freeman , k. c. , quinn , p. j. , 1997 apj 490 , 143 bland - hawthorn , j. ; jones , d. h. ; 1998 pasa 15 , 44 bland - hawthorn , j. ; 1998 aspc 136 , 113 bland - hawthorn , j. , vlajic , m. , freeman , k. c. , draine , b. t. , 2005 apj 629 , 239 bochkarev , n. g. ; sunyaev , r. a. ; 1977 sva 21 , 542 boissier , s. , 2007a iaus 235 , 303 boissier , s. , et al . , 2007b apjs 173 , 524 boissier , s. , gil de paz , a. ; boselli , a. ; buat , v. ; madore , b. ; chemin , l. ; balkowski , c. ; amram , p. ; carignan , c. ; van driel , w. ; 2008 apj 681 , 244 briggs , f. h. , 1990 apj 352 , 15 christlein , d. , zaritsky , d. , 2008 apj 680 , 1053 christlein , d. ; bland - hawthorn , j. , 2008 arxiv 0808.3878 cepa , j. ; 1998 ap&ss 263 , 369 dieter , n. h. , 1962 aj 67 , 217 erwin , p. ; pohlen , m. ; beckman , j. e. , 2008 aj 135 , 20 erwin , p. ; pohlen , m. ; gutierrez , l. ; beckman , j. e. , 2007 arxiv0712.1473 ferguson , a. m. n. , wyse , r. f. g. , gallagher , j. s. , hunter , d. a. , 1998 apj 506 , 19 ferguson , a. m. n. , gallagher , j. s. , wyse , r. f. g. , 1998 aj 116 , 673 freeman , k. c. , 1970 apj 160 , 811 herbert - fort , s. ; zaritsky , d. ; moustakas , j. ; christlein , d. ; wilcots , e. ; baruffolo , a. ; di paola , a. ; fontana , a. ; giallongo , e. ; pogge , r. w. ; ragazzoni , r. ; smareglia , r. , 2009 apj 700 , 1977 van de hulst , h. c. , raimond , e. , van woerden , h. , 1957 ban 14,1 kazantzidis , s. , bullock , j.s . , zentner , a.r . , kravtsov , a.v . ,
moustakas , l.a .
, 2008 , apj , 688 , 254 kennicutt , r. , 1989 apj 344 , 685 kroupa , p. , 2007
, astro.ph/0703124 van der kruit , p. c. , 1987 a&a 173 , 59 lauberts , a. ; valentijn , e. a. , 1989 , european southern observatory madsen , g. j. ; reynolds , r. j. ; haffner , l. m. ; tufte , s. l. ; maloney , p. r. , 2001 apjl 560 , 135 maloney , p. ; 1993 apj 414 , 41 martin , c. , kennicutt , r. , 2001 apj 555 , 301 meurer , gerhardt r. ; wong , o. i. ; kim , j. h. ; hanish , d. j. ; heckman , t. m. ; werk , j. ; bland - hawthorn , j. ; dopita , m. a. ; zwaan , m. a. ; koribalski , b. ; seibert , m. ; thilker , d. a. ; ferguson , h. c. ; webster , r. l. ; putman , m. e. ; knezek , p. m. ; doyle , m. t. ; drinkwater , m. j. ; hoopes , c. g. ; kilborn , v. a. ; meyer , m. ; ryan - weber , e. v. ; smith , r. c. ; staveley - smith , l. ; 2009 arxiv 0902.0384 moustakas , j. ; kennicutt , r .- c . ,
jr . 2006 , apjs , 164 , 81 pohlen , m. , dettmar , r .- j . , ltticke , r. , aronica , g. , 2002 a&a 392 , 807 pohlen , m. , trujillo , i. , 2006 , aa , 454 , 759 roskar , rok ; debattista , victor p. ; quinn , thomas r. ; stinson , gregory s. ; wadsley , j. ; 2008 apjl 684 , 79 sanchez - saavedra , m. l. , battaner , e. , guijarro , a. , lopez - correidora , m. , castro - rodrguez , n. , 2003 a&a 399 , 457 sellwood , j. a. ; binney , j. j. ; 2002 mnras 336 , 785 trujillo , i. , pohlen , m. , 2005 apjl 630 , 17 thilker , d. a. , et al . ,
2005 apj 619 , 79 thilker , d. a. , et al .
2007 apjs 173 , 538 de vaucouleurs , g. , de vaucouleurs , a. , corwin jr . ,
buta , r. j. paturel , g. , fouque , p. , 3rd reference catalogue of bright galaxies v3.9 vlajc , m. ; bland - hawthorn , j. ; freeman , k. c. ; 2009arxiv0903.185 vogt , n. p. ; haynes , m. p. ; herter , t. ; giovanelli , r. ; 2004 aj 127 , 3273 williams , b.f . , dalcanton , j.j . , dolphin , a.e . ,
holtzman , j. , and sarajedini , a. 2009 apjl 695 , 15 zaritsky , d. , christlein , d. , 2007 aj 134,135 | the surface brightness profile of h@xmath0 emission in galaxies is generally thought to be confined by a sharp truncation , sometimes speculated to coincide with a star formation threshold . over the past years
, observational evidence for both old and young stellar populations , as well as individual h ii regions , has demonstrated that the outer disk is an actively evolving part of a galaxy . to provide constraints on the origin of the aforementioned h@xmath0 truncation and the relation of h@xmath0 emission in the outer disk to the underlying stellar population , we measure the shape of the outer h@xmath0 surface brightness profile of 15 isolated , edge - on late - type disk galaxies using deep , long - slit spectroscopy .
tracing h@xmath0 emission up to 50% beyond the optical radius , @xmath1 , we find a composite h@xmath0 surface brightness profile , well described by a broken - exponential law , that drops more steeply in the outer disk , but which is not truncated . the stellar continuum and h@xmath0 surface brightness both exhibit a break at @xmath2 , but the h@xmath0 emission drops more steeply than the stellar continuum beyond that break .
although profiles with truncations or single exponential laws correctly describe the h@xmath0 surface brightness profiles of some individual galaxies , flexible broken - exponentials are required in most cases and are therefore the more appropriate generic description .
the common existence of a significant second surface brightness component beyond the h@xmath0 break radius disfavors the hypothesis that this break is a purely stochastic effect .
galaxies : kinematics and dynamics galaxies : formation galaxies : evolution galaxies : spiral galaxies : structure |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.